Flowfield Measurements in a Transonic Turbine by Emmanuel Michel d'Hoop Ingenieur, Ecole Centrale Paris 1993 Submitted to the Department of Aeronautics and Astronautics in partial fulfillment of the requirements for the degree of Master of Science at the MASSACHUSETTS INSTITUTE OF TECHNOLOGY May 1995 © Massachusetts Institute of Technology 1995. All rights reserved. Author .......................................... . .... Department of Aeronautics and Astronautics May 22, 1995 A7A/ Certified by ..... .- - . .. Accepted by .................. ......... Alan H. Epstein Professor Thesis Supervisor ... Professor Harold Y. Wachman Chairman, Department Graduate Committee MASSACHUSETTS INSTITUTE 'JUL 07 1995 LIBRARES AerO Flowfield Measurements in a Transonic Turbine by Emmanuel Michel d'Hoop Submitted to the Department of Aeronautics and Astronautics on May 22, 1995, in partial fulfillment of the requirements for the degree of Master of Science Abstract This work presents experimental flow field investigations conducted on the Blowdown Turbine, a transient transonic turbine test rig capable of simulating engine-like flow conditions by scaling and matching of relevant non-dimensional parameters. This test rig is equipped for heat transfer measurements with inlet temperature distortions. The influence of inlet turbulence on rotor heat transfer was studied using a turbulence generating grid located upstream of the turbine stage. Hot wire anemometry was used to measure the turbulence intensity and frequency content and compare them with real engine data. To produce narrow and high temperature hot streaks and simulate extreme combustor temperature distortions, long hot gas injectors were designed, calibrated and tested on the rig. The resulting hot spots were narrower than the ones obtained with a previous injector design, but their peak temperature was slightly lower, indicating a mixing rate higher than expected. Finally, a Particle Image Velocimetry (PIV) system has been set-up to perform velocity measurements in the rotor passage of the turbine. The flow is seeded with small particles (0.51im), which is required for accurate velocity measurements in a transonic flow. The system uses a double pulsed Nd-YAG laser and a video camera. The pulsed laser provides the optical power output necessary to visualise submicron particles. The optical system has been designed to meet the requirements imposed by operations on a transient rig like the Blowdown Turbine, particularly in terms of tunnel vibrations. The data reduction algorithm used to process the PIV images is a direct interrogation method, as opposed to 2D Fourier transform techniques. The direct approach is better suited to high speed flow PIV which produces images with a low data density. The accuracy obtained for the velocity measurements is 20% with the video system. This is not sufficient for flow analysis but can be improved by moving to a film camera which has a much higher resolution and has already been tested successfully. Thesis Supervisor: Alan H. Epstein Title: Professor Acknowledgments I would like to thank Prof. Alan H. Epstein and Dr. Gerald R. Guenette for their guidance and support throughout the course of this work, and for the opportunity to work with them on the Blowdown Turbine facility. Special thanks to Dr. Tonghuo Shang for his precious advice with experimental matters, and his help in running the tunnel. Many thanks also to Prof. Peter J. Bryanston-Cross from Warwick University, without whom PIV on the Blowdown Turbine would never have been a reality. I would like to thank the other students who worked with me on this project. Ismail Ceyhan for designing the window assembly and measuring the tunnel movement, and Daciana Udrea for setting up the AP software on the GTL workstations. Thanks to Viktor Dubrowski, Bill Ames, and James Letendre for their help in turning theory into practice and making things work, Holly Rathbun and Robin Courchesne for taking care of all the administrative details, and Diana Park for the illustrations. Funding for my stay at MIT was provided in great part by SNECMA. I wish to thank them, and particularly the people of the turbine division, for making this possible. Thanks also to Prof. Sebastien Candel of the Ecole Centrale Paris for his help and guidance along the road that lead me to MIT. Finally, I would like to thank all the people who helped make my stay in Boston an enjoyable one. The ever-growing MIT French Connection for making me feel a little at home, Diego, Roland and Franckie for many late night beers, Vincent for the introduction to the subtleties of the French Canadian language, the Aero Astro Hockey team for the fun and excitement, and all the others (the list would be too long). Contents 1 Introduction 10 2 Experimental Facility 13 2.1 Blowdown Turbine wind tunnel ..................... 13 2.2 Temperature Distortion Generators . .................. 15 2.2.1 Radial Temperature Distortion 15 2.2.2 Circumferential Temperature Distortion . ........... 16 Instrumentation and Data Acquisition Systems . ........... 17 2.3 3 2.3.1 Temperature sensors ....................... 17 2.3.2 Pressure sensors .......................... 18 2.3.3 Data Acquisition Systems 18 .................... Turbulence Measurements in the Blowdown Turbine 27 3.1 28 3.2 3.3 4 . ................ Experimental procedure ......................... 3.1.1 Instruments and calibration 3.1.2 Temperature fluctuations ..................... ................... 28 30 Results and data analysis ......................... 3.2.1 Turbulence Intensity ..................... 3.2.2 Power Spectral Density .................... Conclusion .................. 31 .. ........ ..... 31 .. 32 .... 33 Characterisation of the NGV Flow Field 41 4.1 Streamline curvature calculation ..................... 42 4.2 Boundary Layer calculations ................... .... 43 4.3 5 6 Hot Spot Injectors ................. .......... 44 4.3.1 General Design .......................... 44 4.3.2 Mass Flow Calibration ...................... 44 4.4 NGV Exit Flow Temperature 4.5 Conclusion ................................. ...................... 47 48 Particle Image Velocimetry 74 5.1 Introduction ................................ 74 5.2 PIV on the Blowdown Turbine . .................. .. ......... ... 75 ..... 75 5.2.1 Seeding 5.2.2 Light Source ............................ 77 5.2.3 Optical System .......................... 79 5.2.4 Imaging system .................. 5.2.5 Rig Movement ........................... 82 5.2.6 System Synchronisation 83 .... .. .... ... .. . ..... . ..................... 81 PIV Results and Data Analysis 103 6.1 Data Reduction and Calibration ..................... 103 6.1.1 The AutoPIV Program ...................... 103 6.1.2 Image Magnification 6.2 6.3 Results . .. .. . . .. .. ....................... ... .. .. .. . . . . . .. .. .. 107 . . .... 108 6.2.1 Timing ............................... 108 6.2.2 Particle image size ........................ 108 6.2.3 M agnification ........................... 109 6.2.4 Seeding Density .......................... 109 6.2.5 Preliminary Data Analysis .................. Conclusion ..................... . . ........... 110 . 111 List of Figures 2-1 General view of the Blowdown Turbine test rig. . ............ 20 2-2 Cross section of the flow path upstream of the turbine stage...... 21 2-3 Cross section of the flow path downstream of the turbine stage 22 2-4 Radial temperature distortion generator. . ........ 2-5 Hot gas injection system (short injector design). . ............ 24 2-6 Instrumentation of the Blowdown Turbine flow path. . ......... 25 2-7 Location of NGV exit static pressure taps. . . ............... 26 3-1 Grid geometry for the Blowdown Turbine. The grid is placed at the ... ....... 23 boundary layer bleed lip .......................... 34 3-2 Power spectral density, nozzle case NGV location. . ........ 3-3 Power spectral density, NGV location. 3-4 Location of turbulence grid and hot wire probe in the Blowdown Tur- . . . ................. bine flow path .................... 35 35 .......... 36 3-5 Circumferential location of the hot wire probe (looking downstream) . 37 3-6 Decrease in pu during a blowdown test. . ................. 38 3-7 Power spectral density of tunnel turbulence without turbulence grid (test 186). Solid line : wire 1, dashed line : wire 2. 3-8 . .......... Power spectral density of tunnel turbulence with turbulence grid (test 187). Comparison with previously measured turbulence data. Solid line : wire 1, dashed line : wire 2, * : measured, o : Moss and Oldfield. 4-1 39 40 Streamline curvature calculation grid. Extended geometry with axial exit flow .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 50 ....................... 4-2 Streamline curvature result. 4-3 Static pressure contours (contour interval : 170 Pa) .......... 4-4 inner and outer radius free stream velocity . .............. 4-5 detail of bleed leading edge free stream velocity 4-6 inner radius displacement thickness for low Re . ............ 4-7 inner radius displacement thickness for high Re 4-8 inner radius momentum thickness for low Re . ............. 56 4-9 inner radius momentum thickness for high Re . ............ 57 4-10 outer radius displacement thickness for low Re . ............ 58 4-11 outer radius displacement thickness for high Re . ........... 59 4-12 outer radius momentum thickness for low Re . ............. 60 4-13 outer radius momentum thickness for high Re . ............ 61 52 53 . ........... 54 55 . ........... 62 .. 4-14 Long hot streak injector desihn. . .................. 4-15 Long injector design. Detail of the nozzle. 51 . 63 . ............... 4-16 Injector mass flow calibration curve. Solid line : theory, o : baseline calibration, x : high pressure, * and + : high temperature. ...... . 64 . 65 4-17 Injector mass flow calibration, deviation from theory. o : baseline calibration, x : high pressure, * and + : high temperature. ...... 4-18 Test 181 NGV exit total temperature survey. Downstream rake sensors 1 through 8. NGV spacing is 100. 66 .................... 4-19 Test 191 NGV exit total temperature survey. Downstream rake sensors 1 through 8 (except 5) ................... ....... 67 4-20 Test 192 NGV exit total temperature survey. Downstream rake sensors 1 through 8 (except 5) ................... ....... 68 4-21 Test 196 NGV exit total temperature survey. Downstream rake sensors 1 through 8 (except 5) ................... ....... 69 4-22 Comparison of NGV exit temperature at 49% span for large and small injectors .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 4-23 Comparison of NGV exit temperature at 14% span for large and small injectors .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 4-24 Comparison of NGV exit temperature at 49% span for small injectors with different mass flows .......................... 72 4-25 Comparison of NGV exit temperature at 14% span for small injectors with different mass flows .......................... 73 5-1 Typical set-up for a PIV experiment. . .................. 85 5-2 Particle response to an oblique shock wave, for different sizes .... 86 5-3 Scatter cross section of spherical particle as a function of diameter and wavelength. ................................ 5-4 86 Seperation between the two Q-switch triggers from the laser sync box. The interval is set to 600 ns ......................... 5-5 87 Seperation between the two laser pulses, detected by a light sensitive diode. .................................... 88 5-6 PIV optical setup ............... 5-7 Setup for PIV with the rotor, using two periscopes and transparent vanes . . . . . . . . . . . . . . . . . . . . ............... 89 . . . . . . . . . . . . .. . 5-8 Light path setup from laser head to test section. . ............ 5-9 Illumination periscope with sheet forming optics. 90 91 . ........ . . 92 5-10 Relative position of probe, laser sheet and test section. View 1. . . 93 5-11 Relative position of probe, laser sheet and test section. View 2. . . . 93 5-12 Viewing window. ................... .......... 5-13 Window frame ........................ ....... 5-14 W indow holder ...................... .......... 5-15 Main test section assembly frame. View 1. .......... 94 95 96 . . . . . 97 5-16 Main test section assembly frame. View 2. . ............... 98 5-17 Main test section assembly frame. View 3. . ............... 99 5-18 Synchronisation of laser pulse and video camera to obtain non interlaced images. .................. ............ 100 5-19 Lag between the camera asynchronous reset signal and the first pulse of the TTL burst. The lag (chl to ch2) is 5.2 ms . ........... 101 5-20 Component synchronisation layout . 6-1 . . . . . . . . . . . . . . . . . . Comparison of highest intensity pixel and bounding box methods for 112 finding the particle center. ......................... 6-2 Gaussian approximation of the light intensity function of the particle im age. .. 6-3 102 . . . . . .. ..... .. .. .. .. . .. .. .. ...... 112 Examples of particle images that can be processed by the Gaussian fit method : noisy profile (a), saturated profile (b), incomplete shape (c), overlapped images (d), interlaced image (e) or distorted image (f). . . 113 6-4 Horizontal line from a ruler picture . . . . . . . . . . . . . . . . . . . 6-5 Fourier transform of the horizontal line light f.unction. .......... 6-6 Raw PIV image, test 204 image#3. ...... 114 115 S . . . . . . . . . . . . . 116 6-7 Raw PIV image, test 204 image#4..... . . . . . . . . . . . . . . 117 6-8 Raw PIV image, test 205 image#3 ...... . . . . . . . . . . . . . . 118 6-9 . . . . . . . . . . . . . . 119 Raw PIV image, test 205 image#4 ...... 6-10 AP processed PIV image, test 204 image#3. S . . . . . . . . . . . . . 120 6-11 AP processed PIV image, test 204 image#4. S . . . . . . . . . . . . . 121 6-12 AP processed PIV image, test 205 image#3. S . . . . . . . . . . . . . 122 6-13 AP processed PIV image, test 205 image#4. S . . . . . . . . . . . . . 123 6-14 NGV exit Mach number, test 204....... . . . . . . . . . . . . . . 124 6-15 NGV exit Mach number, test 205....... . . . . . . . . . . . . . . 125 6-16 Laser pulse and frame grabbing sync signals, ttest 204. . ......... 126 6-17 Laser pulse and frame grabbing sync signals, I;est 205. . ......... 127 6-18 Particle image light intensity function, test 204. ............. 128 6-19 Particle image contour, test 204 . . . . . . . . . . . . . . . . . . . . . 129 6-20 Particle image light intensity function, test 20 5. . . . . . . . . . . . . 130 6-21 Particle image contour, test 205 . . . . . . . S . . . . . . . . . . . . . 131 6-22 Particle pair from image#3, test 204 ..... S. . . . . . . . . . . . . 6-23 Particle pair from image#3, test 205 ...... S. . . . . . . 132 . . . . . 133 Chapter 1 Introduction In modern aircraft gas turbine engines, the turbine inlet temperature is one of the determining factors of general performance. The specific thrust, and overall efficiency of the engine are improved when this temperature is increased [3]. This provides a strong incentive to operate turbines at temperatures as high as possible, which means as high as the blade alloy will withstand. Improvements in materials over the years have permitted continual small increases in turbine inlet temperatures, up to the 1000-1200K range [3] [4]. The introduction of turbine blade cooling in the 1960's allowed much higher temperatures, up to 1650K in modern commercial engines, and even higher in military engines. However, these are still far from the stoichiometric flame temperature of 2300-2500K, which means there is still a significant possibility for improvement in turbine operating temperatures, and engine efficiency. Cooling techniques consist in bleeding air from the compressor, carrying it to the turbine and injecting it through the blades. Several methods are used to achieve this function [3] [4] [5]. With convection or impingement cooling, the air flows inside the blade and decreases its temperature by convective heat transfer, whereas with film or transpiration cooling, air is injected through holes in the blade into the boundary layer in order to decrease the heat transfer between the free stream (hot) flow and the blade. Film cooling is the most widely used technique in modern engines. Although more disruptive to the flow than internal cooling, it allows the highest operating temperature. Transpiration cooling performs better : it is less disruptive because the holes are small relative to the boundary layer thickness. However, it is more difficult to implement in practice and is not yet used in commercial engines [3] [5]. The amount of cooling applied on the different sections of a blade geometry must be adequately matched with the heat load. On the one hand, insufficient cooling will result in excessive material temperature and a considerable reduction of the turbine life, or even blade failure in extreme cases. On the other hand, too much cooling at some point of the blade surface will cause a strong temperature gradient in the material and lead to thermal stress. In order to implement an efficient cooling scheme in the design of a turbine blade, understanding of the flow phenomena which affect heat transfer is necessary. These include unsteady effects caused by wakes and shocks in transonic turbines, and inlet temperature distortions generated by the combustor, which have a strong influence on both peak and averaged flow variables as shown by Shang [1]. The purpose of this study is to help investigate those phenomena and understand how they affect turbine heat transfer and aerodynamic performance. All the experimental work presented here has been conducted on the MIT Blowdown Turbine facility, a fully scaled transonic transient test rig which is described in details in chapter 2. It has been modified to generate both radial and circumferential temperature distortions at the inlet of the turbine stage in order to study their influence on the rotor blade heat transfer coefficient [1]. The radial distortion is created by an unevenly heated grid matrix, while the circumferential distortion (also called hot streaks) is created by a system of hot gas injectors. To aide in analysing the heat transfer data, a series of flowfield measurements and simulations were also conducted and are presented in this thesis. A turbulence generating grid was used during some of the tests to investigate the influence of inlet flow turbulence on the heat transfer coefficient. This grid had been characterised on a test bench outside the blowdown turbine, but not in the fully scaled rig conditions. Chapter 3 describes turbulence measurements made directly in the rig during blowdown tests to characterise the tunnel turbulence level with and without the grid and correlate it with the heat transfer data. Chapter 4 discusses a modified design for the hot gas injectors, the goal of which is to create narrower hot streaks, more representative of real engine conditions. Several measurements and computational simulations were made to characterise the turbine inlet flow with the new injector design. The subsequent chapters present the implementation of a PIV (Particle Image Velocimetry) system to perform 2D velocity measurements in the turbine stage of the facility. Chapter 5 presents the PIV technique in general, and describes the experimental setup used to implement this technique on the Blowdown Turbine test rig. Chapter 6 describes the method used to process the raw images obtained from the tunnel and get velocity and flow angle data. The first PIV results obtained from the Blowdown Turbine are also presented and analysed in this chapter. Chapter 2 Experimental Facility This chapter describes the test rig on which the experimental work presented in this thesis has been conducted. The wind tunnel and its operating principle, the instrumentation, and the data acquisition systems are presented here. 2.1 Blowdown Turbine wind tunnel The Blowdown Turbine is a fully scaled transient wind tunnel capable of fully simulating the non dimensional flow conditions for modern transonic axial turbines. Table 2.1 shows the scaling of the rig and the values of the main parameters compared to those of a turbine operating in an engine environment. All the non-dimensional parameters relevant to the flow physics are simulated. The usable test time is approximately 300 ms which is large compared to the characteristic time scale of the flow and to the rotor blade passing frequency. Thus the turbine operates in a quasi steady-state mode, allowing the investigation of unsteady phenomena. The ratio of specific heats, 7, is reproduced using a mixture of gases (argon and freon 12 or argon and C02). However, air can also be used for tests with the Nozzle Guide Vanes only, for which the value of' is no as critical. Because of their thermal inertia, the blades and rig walls remain at constant temperature (i.e. room temperature) during the short test time. This way, the desired Full scale Air Working fluid Ratio of specific heats 1.28 Mean metal temperature 1118 0 K Gas/wall temperature ratio, Tg/Tm, 1.59 Inlet total temperature, T 17800 K True NGV cord 8.0 cm 2.7 x 106 Reynolds number Inlet total pressure 19.6 atm Outlet total pressure 4.5 atm Prandtl number 0.752 12,734 Rotor speed, rpm Mass flow, kg/s 49.00 MIT Blowdown Argon-Freon 12 1.28 295 0 K (room temp) 1.59 4780 K 5.9 cm 2.7 x 106 4.3 atm 1.0 atm 0.755 6,190 16.55 Power, W 24.88 x 106 1.078 x 106 Test time continuous 0.3 sec Table 2.1: MIT Blowdown Turbine scaling gas to wall temperature ratio can be obtained using inflow temperatures lower than would be required in a continuous wind tunnel. Also, because the fluid is cooler, and has a higher molecular weight, a lower rotor speed is needed to get the same tip Mach number. All this makes the facility easier and safer to operate. A schematic of the facility is presented in figure 2-1. The main components are the supply tank, fast acting valve, temperature distortion generators, upstream and downstream translators, test section and dump tank. Cross section views of the flow path between the valve and the dump tank are presented in figures 2-2 and 2-3. The annulus is divided in three sections as shown in figure 3-5 of chapter 3. The turbine stage located in the test section is a 4 to 1 pressure ration transonic turbine designed by Rolls Royce and referred to as ACE (Advance Core Engine) stage. The geometric scaling factor is approximately 3/4. Table 2.2 gives the dimensions of the turbine stage. The flow path upstream of the test section includes a contraction in front of the NGV inlet which simulate an engine geometry (combustor exit). Also, boundary layer bleeds are placed at the entrance of the contraction to capture the boundary layer from the upstream flow and ensures that a relatively clean flow enters the turbine rotor NGV Number of airfoils 36 61 Mid-span axial cord 3.4 cm 2.6 cm Inlet angle Exit angle Axial gap Rotor tip diameter 570 -650 00 740 1.1 cm 55 cm Table 2.2: Dimensions of ACE stage stage. Before a test, the tunnel is pumped down to vacuum and the supply tank and temperature distortion generators are heated to the desired temperatures. Then the valve is closed and the supply tank is filled with the test gas mixture to a pressure of 4 to 6 atm, depending on the test conditions. During a rotor test, the rotor is then spun up to above the pre-selected design speed, decelerates slowly due to friction forces, and triggers the valve when it reaches the design speed. During an NGV only test, the system is triggered manually. The valve opens 30 to 50 ms after the trigger. The flow takes about 200 ms to establish, and then the blade passages and throttle orifice remain choked for approximately 300 ms, which is the useful test time. The trigger also starts the eddy current brake (if the rotor is used) and the data acquisition systems. More details about the design and operation of the facility can be found in references [6] and [7]. 2.2 Temperature Distortion Generators Two types of temperature distortions can affect the performance of a turbine : radial and circumferential distortions. Both can be simulated in the Blowdown Turbine. 2.2.1 Radial Temperature Distortion The radial temperature distortion generator is an annular stainless steel honeycomb matrix which acts as a heat exchanger and introduces a radial temperature profile into the flow. Details of its design can be found in [8]. Its temperature is controlled by electrical heater wires at the center, and by jacketed walls on its inner and outer diameters in which water or oil can be fed to heat or cool the matrix boundary (see fig 2-4). The honeycomb pattern and length are designed so that the fluid going through it is heated to the metal temperature. This way, the inlet flow temperature profile is the profile measured on the matrix. A set of 30 type J thermocouples is used to monitor the grid temperature during heat-up. The matrix is electrically heated in its center and the heat diffuses toward the edges, creating a radial temperature profile. The thermal conductivity of the matrix is much higher in the circumferential direction than in the radial direction, and thus the radial profile is constant along the whole circumference of the annulus [1. Because of particle contamination from the grid, a 20 micron diameter particle filter is used downstream of the RTDG. 2.2.2 Circumferential Temperature Distortion Because its thermal conductivity of the matrix is too high in the circumferential direction, the heater grid cannot be used to create hot streaks in the flow. Four hot gas injectors located in sector A of the wind tunnel are used for this purpose. As shown in figure 2-5, the hot gas injection system takes advantage of the pressure drop across the grid and filter to divert a small percentage of the flow, heat it up and re inject it in the main flow at the inlet of the turbine stage. The bypass flow is heated by a 380 mm long tube bundle heat exchanger. The diameter of the injection hole is 29 mm. Further details of the gas injector design and calibration can be found in [1]. To create narrower and sharper hot spots, an alternative design for the injectors was made as shown in figure 4-14 of chapter 4. The injection hole diameter is now 12.7 mm and the injector extends along the annulus up to 12.7 mm from the NGV leading edge. The hot streaks created here are thus narrower and closer to the turbine stage inlet. 2.3 Instrumentation and Data Acquisition Systems The instrumentation of the Blowdown Turbine includes static and translating temperature and pressure sensors and heat flux gauges on 3 rotor blades. The instrumented blades for heat transfer measurements are thoroughly documented in references [1] and [9] and will not be described here in further detail. This section will summarise the information relating to the instrumentation used during the presented research. Figure 2-6 shows the complete flow path of the tunnel with the location of the temperature and pressure sensors. Translating rakes are used upstream and downstream of the turbine stage for both pressure and temperature measurements to get a 2D picture of the flow. Figure 2-6 shows the rig configuration with the rotor installed. When the rotor is removed for NGV only testing, the downstream translator is placed directly at the NGV exit plane, 27 mm away from the trailing edge. Details about the design and operation of the translators can be found in [1]. 2.3.1 Temperature sensors The temperature sensors are type K thermocouples located on the upstream and downstream translators and directly behind each hot gas injector to measure the peak temperature of the hot spots. The reference junction for these thermocouples are located in a Dewar container filled with silicon oil and mounted outside of the test section. The very high heat capacity of the oil provides a constant reference temperature. The upstream translator has 11 gauges, the downstream 8. In addition, total temperature sensors are located on the stationary rake in front of the boundary layer bleed lip. For more detailed information about the sensors and their calibration and operation, see reference [2]. 2.3.2 Pressure sensors Pressure instrumentation includes total pressure sensors on both translators (3 upstream and 8 downstream), 1 total pressure sensor on the stationary rake at the test section inlet, total pressure sensors in the injector flow paths, and wall static pressure sensors and the NGV exit and downstream of the stage. The injector total pressure taps are used to measure the mass flow through the hot gas injection system as shown in ref [1]. Static pressure sensors are located at the NGV exit on the inner and outer wall of the annulus (see fig 2-7). Their angular position is at the center of sector C. 2.3.3 Data Acquisition Systems Two A/D systems are used to acquire data from the sensors : The MIT Blowdown Turbine 12 bit A/D system has 8 multiplexers with 16 channels each. Data from the high frequency total pressure rake sensors are sampled using this system. The lower frequency pressure sensors and other additional transducers such as hot wire probes are sampled using the multiplexers. This system is run by a DEC micro VAX II computer. The sampling frequency varies as follows with time : * 0-250 ms at 50 kHz * 250-550 ms at 200 kHz * 550-1200 ms at 50 kHz * 1200 ms-10 min at 50 Hz An Analogic model HSDAS-16 16 bit A/D card with an AMUX-64-16 multiplexer provides 64 channels multiplexed into a single high speed A/D converter. It is installed on a Dell 486D-50 IBM PC compatible computer. The upstream and downstream temperature probes, as well as the injector temperature probes are sampled using this A/D. The sampling frequency varies as follows : * 0-1000 ms at 3.03 kHz * 1 s-10 min at 1 Hz An additional ADTEK Corp. model AD 830 12 bit A/D system run on a Dell 486 EISA IBM PC compatible computer was used to provide extra high speed channel for the rotor heat flux gauges when testing with the instrumented rotor. This A/D was synchronised to the MIT BDT A/D system with the same clock and sampling rate. Finally, the IBM PC used to control the translators also recorded the translator positions. 1 Meter Oq Injector C3 I-l O 0 e40 (1 C,., v:: (1) - 1I 0- c d3 It CP.- C4- cn CA C- p d. IHeaters Flow -' ' Oil Jacket 0.H Side Sie(oknoontem SS Honeycomb Center (Looking Downstream) External Heating Tape 01 C4+ O) P) D Co Co Upstream Translator I.A. (YQ Injector Supply Pressure Iap Sac Static Tap 0 1.00" 1.00" 0.10" Tip Hub Figure 2-7: Location of NGV exit static pressure taps. Chapter 3 Turbulence Measurements in the Blowdown Turbine The influence of turbulence on rotor heat transfer was explored by operating the facility with high and low levels of turbulence. In order to determine this influence accurately, it is necessary to know the intensity and frequency content of the turbulence generated in the tunnel. Previously, the intensity of the turbulence naturally produced by the Blowdown Turbine tunnel had been measured as less than 1% before the contraction into the NGV's. Since then the facility was modified by the addition of a distortion generator and in-line filter, downstream of which a turbulence grid was added. The turbulence grid has been designed to produce turbulence with similar characteristics (intensity and frequency content) as the one encountered under engine conditions. The basis used for that design was the work of Moss and Oldfield [10]. The grid geometry is based on 0.25" square bars on 1" centers (Fig 3-1). The characteristics of this turbulence grid have been measured in a low speed wind tunnel with a geometry similar to that of the turbine inlet. The data show that the grid generates a turbulence spectrum that agrees well with the results obtained from Moss and Oldfield, but with a somewhat lower intensity, indicating that there is less energy contained in the turbulence generated by the grid than in the engine (Fig 3-2). The measured grid turbulence was in agreement with that given in the literature (Fig 3-3). This chapter presents the results of turbulence measurements made in the Blowdown Turbine rig in front of the NGV leading edge, in the contraction (Fig 3-4 and 3-5). To determine the influence of the grid on the level of turbulence in the tunnel, several measurements were made with and without the grid. 3.1 3.1.1 Experimental procedure Instruments and calibration Hot wire anemometry has been used to measure the characteristics of the turbulence in the Blowdown turbine tunnel. A Dantec 55P71 dual hot wire probe was used with a Dantec 56C01 constant temperature anemometer. Because they can drift rapidly, hot wires usually have to be calibrated a short time before data aquisition, and preferably under the actual testing conditions. The Blowdown Turbine is a short duration facility using various gas mixtures, so that a conventional calibration is difficult here. The testing conditions are not easily reproducible in test bench. Also, tests last only 300 ms. However, during this short time all the mean aerodynamic quantities such as total pressure and total temperature can be measured by other tunnel instruments and all the non-dimensional numbers are known. It is therefore possible to do an in situ calibration of the hot wire : the mean hot wire voltage is used with the other mean aerodynamic quantities to calculate the calibration constants, and then these constants are used to convert the AC part of the hot wire signal into velocity fluctuations. The hot wire voltage varies with the fluid velocity and temperature according to the following equation : V 2 = (X + Y (/-)(T, - TW) derived from King's law [11], where : V = hot wire voltage X, Y = hot wire calibration constants p = fluid density (3.1) u = fluid velocity T, = hot wire temperature T,. = fluid temperature The calibration is made between two points of the data set, using averaged values. We get : X + Y/u = X + Y/ = - V2 V2 2 TW - T0, (3.2) which gives : Y X 2T,, T, -T, = V T - Too, - Y (3.3) F for the considered hot wire. This type of calibration is possible because there is a significant decay in density (and thus in the value of pu) during the blowdown (Fig 3-6). Knowing X and Y, the slope of the curve pu = f(V) can be calculated from equation 3.1 : V pu = - T V T 2 x 2 - X (3.4) (3.4) the slope is : S(pu) aV 4V Y2(T - T) V2 T- (3.5) Too Using this slope, the variations in pu can be calculated from the variations in hot wire voltage : S= P (3.6) 3.1.2 Temperature fluctuations During a blowdown test with temperature distortions, the temperature at the NGV inlet varies significantly with time, with an amplitude of about 100C. The hot wire signal depends on the velocity but also on the temperature of the fluid (eq 3.1). Temperature variations will therefore interfere with the turbulence measurements and must be taken into account. It was estimated that a 100 C amplitude oscillation in the NGV inlet temperature could cause a fluctuation in the hot wire signal of the same amplitude as the one induced by the turbulence, and thus cause an unacceptable error in the measurements. To compensate for this, it is necessary to measure the temperature fluctuations at the location of the hot wire. By correlating the signals from the two wires, it is possible to deduce this temperature, but the precision obtained is too low to allow accurate turbulence measurements. During tests without temperature distortions, the temperature of the flow has much less variations (about 10C) and can be assumed constant. The turbulence measurements presented further in this report were taken during tests without temperature distortions. If T, is constant then eq 3.3 becomes : V2 - V2 X T V2 TOO - YVPU replacing in eq 3.5 and using pu = PM 4MV S= 2 1 Y ( V22 - V (2 (3.7) T we obtain + P2 2- +P V (3.8) 3.2 3.2.1 Results and data analysis Turbulence Intensity The turbulence intensity is defined as : Tu- U U (3.9) Equation 3.8 shows that the slope a has a single dependence in fluid temperature with the term . Since p = , and u' = , u' has a single V7AT factor. U is also a direct function of f',RT so this term cancels out in the calculation of Tu = U. . Thus the data reduction method to obtain the turbulence intensity does not depend on the fluid temperature. (This is true if the fluid temperature is assumed constant, which is the case here). Tu depends only on P, and on the values of the pressure and hot wire voltage at the two points chosen for calibration. The pressure dependence is very small and the error in Tu due to errors in pressure measurements is negligible. Thus the main source of error in the estimation of Tu comes from the calibration, more particularly from the choice of the two calibration points. Table 3.1 shows the values obtained for Tu during a blowdown test made without the turbulence grid. The different points correspond to results obtained from two different hot wires at the same location. The data sets from the two hot wire were processed using the method explained above, and the data reduction was repeated with different calibration intervals (i.e. different "1" and "2" points in the data set). This gives an estimate of the range of the error due to calibration uncertainties. The calibration intervals used range from 235-485 ms to 265-515 ms by increments of 5 ms. In addition to the uncertainty due to calibration errors, there is a constant difference between the values given by the two wires which cannot be explained. The value of Tu in the tunnel without the turbulence grid ranges between 2.5% and 2.9%. Table 3.2 gives the same results for a blowdown test made with the turbulence grid. The same observations as before can be made. The difference between the two wires is higher in this case. The turbulence intensity in the tunnel with the turbulence calibration window wirel wire2 1 2 3 4 5 6 7 2.84 2.78 2.78 2.77 2.67 2.68 2.67 2.64 2.58 2.59 2.58 2.52 2.53 2.51 Table 3.1: Turbulence intensity (%) without the grid. calibration window 1 2 3 4 5 6 7 wirel 9.90 10.06 10.00 8.89 8.35 8.38 8.15 wire2 7.97 8.19 8.27 7.58 7.29 7.25 7.02 Table 3.2: Turbulence intensity (%) with the grid. grid ranges between 7% and 10%. 3.2.2 Power Spectral Density The power spectral density distribution is defined as : G(k)= rUNAt IX(f)l (3.10) Where X(f) is the Fourier transform of the u' signal : N-1 X(f) = At E u'(r)e-i2 wfNAt (3.11) n=o Fig 3-7 shows the power spectral density obtained for a blowdown test without the turbulence grid. The spectrum given by wire 1 is slightly above the one given by wire 2, which is consistent with the fact that the turbulence intensity given by wire 1 is higher than the one given by wire 2. Fig 3-8 shows the same results for a test with the turbulence grid. For reference, the design spectrum (i.e. the one measured by Moss and Oldfield [10]) and the spectrum measured in the low speed wind tunnel are also presented. Relative to those two, the blowdown turbulence intensity is higher by several dB up to a wave number of about 80 and drops off more quickly with frequency. 3.3 Conclusion The turbulence in the Blowdown Turbine tunnel has been characterised using hot wire anemometry. The turbulence intensity and frequency content have been measured, with and without the turbulence grid in the tunnel. The turbulence intensity ranges between 2.3% and 3% for a test without the turbulence grid and between 7% and 10% for a test with the grid. Overall, the turbulence created met the program goals of generating turbulence levels and spectra approximating those measured by Moss and Oldfield in an engine-like environment. The grid turbulence exceeds the design goal up to wave numbers of 50-80 and then drops below it. These measurements were taken during tests without temperature distortions and therefore do not account for any possible effect of these distortions on the turbulence. -0.25" A View A View 1.48" 120 ° R=7.52" Figure 3-1: Grid geometry for the Blowdown Turbine. The grid is placed at the boundary layer bleed lip. W'venwnctncr k(1rn) Figure 3-2: Power spectral density, nozzle case NGV location. -35 Wavenumber k(1/m) Figure 3-3: Power spectral density, NGV location. Figure 3-4: Location of turbulence grid and hot wire probe in the Blowdown Turbine flow path. Hot Spot Injectors (4) Translator Rakes 3-5: TTFigure Circumferential location the of wire hot (looking probe downstream) PT Figure 3-5: Circumferential location of the hot wire probe (looking downstream) Decay in ro*u during blowdown 0 100 200 300 400 500 600 700 800 900 time (me) Figure 3-6: Decrease in pu during a blowdown test. 1000 Power Spectral Density Test turbl86 without turbulence grid C. - LU -a 0 6 -5 -1 Wave number k=2*pi*f/U Figure 3-7: Power spectral density of tunnel turbulence without turbulence grid (test 186). Solid line : wire 1, dashed line : wire 2. 39 Power Spectral Density Test turbl 87 with turbulence grid in : ; : " I;. . 0 X ... .......... ......... ...... ... ..... ... ...... ................. ........ -60 . .... .:......... ...... .... -.......... .. . . . 2 ... \ . .'- . . . . . . -701 ...... ..i ... .. . ..... ..... ... ... .. .. . ................. -80 . . .. . ........................ -90 .. S : : : . . . ." -" 2... 1 ................. ......... ! : I flf S111u .vv Wave number k=2*pi*f/U Figure 3-8: Power spectral density of tunnel turbulence with turbulence grid (test 187). Comparison with previously measured turbulence data. Solid line : wire 1, dashed line : wire 2, : measured, o : Moss and Oldfield. 40 Chapter 4 Characterisation of the NGV Flow Field The hot spot injectors used for the rotor heat transfer measurements research program [1] which are described in section 2.2.2 produce hot spots of a fairly large diameter (more than 1") and quite far away from the NGV leading edge (about 14 injector diameters). This leaves the hot fluid jets enough time to mix out significantly, and as a result, the hot spots are about the size of an NGV passage and have a low peak temperature. These hot spots resemble the design intent of an engine, but more extreme spots are found in some combustors and are desired for exercising CFD codes. An alternative design for longer and narrower hot gas injectors was implemented in an attempt to solve this problem. This chapter describes the design and calibration of the new injectors and the aerodynamic measurements and simulations done to characterise the inlet and outlet flow field of the NGV's with the new injectors in use. First, a streamline curvature calculation was performed to obain static pressure and velocity information for the inlet contraction in front of the stage. These data were then used for a boundary layer calculation on both the inner and outer wall of the annulus between the boundary layer bleed lip and the NGV leading edge. The purpose of these simulations is to investigate the boundary layer developing in the contraction downstream of the bleed and to determine its influence on the NGV flow Temperature (K) Gas constant R (J/(kgK)) Cp (J/(kgK)) Cv (J/(kgK)) Specific heat ratio y Viscosity /1 452.0 287.04 1022.0 734.0 1.39 2.49 x 10- 5 Table 4.1: gas properties Inlet static pressure (Pa) Inlet density (kg/m') Inlet Mach number Static pressure ratio 3.99 x 105 3.07 0.07 0.99 Table 4.2: input aerodynamic conditions field. Temperature measurements were made upstream of the stage using fixed total temperature probes in front of the injector nozzles and downstream of the stage using the translating rake which was located at the NGV exit (see section 2.3). 4.1 Streamline curvature calculation The code used for this preliminary calculation is AMISES, a simplified axisymetric version of the MISES 2D code which is described in reference [14]. The geometry of the inlet contraction has been estimated partially from the input file of a former calculation done by Rolls-Royce and partially from the blueprints of the rig itself. The gas properties used for this case were computed using the Blowdown Turbine test gas routine for a temperature of 4500K and are given in table 4.1. The inlet aerodynamic conditions are given in table 4.2. The grid used for this calculation is shown in figure 4-1. It covers the area between the boundary layer bleed and the NGV leading edge. The geometry was extended in order to provide an axial exit flow condition which is required by the code. The outer wall is a straight line and posed no difficulty. The inner wall was extended using a 2" degree polynomial fit. Four computation cells were added to the grid. In other words, the NGV leading edge is actually located four cells before the end of the grid. Case 1 2 3 4 Re = _ L (m) 568759 0.14 (inlet axial length) 568759 0.14 1577637 0.4 (test section diameter) 1577637 0.4 Tripped no yes no yes Table 4.3: input conditions for the different cases Results of the calculation are shown in figures 4-2 (output grid with streamlines) and 4-3 (static pressure contours). The free stream velocities for the boundary layer calculations on the inner and outer walls are obtained from the Mach number, which is derived from the static to total pressure ratio. Results are plotted in figure 4-4. On the outer wall, which is of constant radius, the velocity increases slowly and regularly from around 30m/s to 50m/s. On the inner wall, which has a fairly large radius variation, the velocity first decreases from 30m/s to 23m/s, probably because of an adverse pressure gradient, and then increases substantially in the second half of the contraction, up to about 70m/s. 4.2 Boundary Layer calculations The free stream velocity information obtained from the streamline curvature calculation is used as input for the MISES boundary layer code. The velocity has been modified at the leading edge to reach zero at the stagnation point (figure 4-5). To investigate the effects of turbulence, runs were made with and without tripping of the boundary layer at the leading edge, at two different Reynolds numbers. The cases are sumarised in table 4.3. The results are presented in figures 4-6 to 4-13. There is no significant difference between the cases 1 and 2 (input Reynolds number calculated using the axial length of the annulus) and 3 and 4 (Re calculated using the test section diameter). On the inner wall, the displacement and momentum thicknesses are of the same order of magnitude in both cases. For the non tripped boundary layer, the laminar-turbulent Injector short long Exit nozzle diameter 2.9 cm (1.14 in) 1.27 cm (0.5 in) Distance to NGV leading edge 14 diameters 1 diamemter Table 4.4: Comparison of short and long injector design transition distance decreases with increasing Re (x=0.055 instead of x=0.03). On the outer wall, thrre is no transition at the lower value of Re, which means that the boundary layer stays laminar along the whole annulus (figures 4-10 and 4-12). When it is tripped, the boundary layer is turbulent from the beginning and grows faster than in the non-tripped cases. The tripping has a more significant effect on the low Re boundary layer than on the high Re one. In all cases, the displacement and momentum thicknesses do not exceed 0.5 mm and 0.14 mm respectively. These results show that in all the cases considered (high and low Reynolds number, tripped and non tripped flow), the boundary layer is very thin. Therefore, its influence on the NGV flow field and on the behaviour of the hot streaks will not be significant. 4.3 4.3.1 Hot Spot Injectors General Design The modified hot gas injectors are designed to create narrower and sharper hot streaks by injecting hot fluid into the flow through a smaller nozzle, and closer to the NGV leading edge than the previous design (table 4.4). The design of the modified injectors, shown in fig 4-14 and 4-15, uses the main components of the previous set-up. Only the injector nozzle has been modified : it is longer, and has a smaller exit diameter. Electrical heater wire has also been added to prevent heat loss as the gas flows through the longer pipe. 4.3.2 Mass Flow Calibration The mass flow calibration has been done using the same method as for the short injectors, which is described in [1]. Because of the smaller exit diameter, however, the mass flow is much less than before and the calibrated orifice plate located in the nozzle does not provide enough flow resistance for an accurate calibration. Therefore another perforated plate has been used in the outside pipe of the injection system, downstream of the tube bundle heater. For tests 190 through 195, the new plate was located upstream of the tube heater and the pressure ratio across the whole system was used for calibration. This did not provide a sufficient control of the flow rate and as a result, the injector mass flow was too high (see table 4.6). For test 196, the plate was moved downstream of the heater. Two static pressure taps were used to measure the pressure upstream and downstream of the orifice plate and provide calibration data [13] [12]. Having the calibrated orifice downstream of the heater also provided a better estimate of the flow total temperature since the tube heater was designed so that the fluid exit temperature is the temperature of the metal. The mass flow was measured during the calibration using a Fisher-Porter model 10A3555A rotameter. The relation between the pressure ratio across the orifice and the mass flow is given by: inj N/RJT P1 P 2 f - =A P 2-f (4.1) where P1 and P 2 are the pressures upstream and downstream, and TT is the total temperature upstream of the orifice. A least square fit on the data gives the value of the effective flow area A for each injector. However, a systematic error proportional to P = PI appears in the P2 process. A modified relation containing a corrective term was thus used : ini& P1 rl R1 = A- 2 -1 P2 F P2 -1+ P )kPP ap P (4.2) Figure 4-16 shows calibration data for injector number 4 taken at room conditions, as well as data taken for the same injector at higher pressure and temperature to check the validity of the calibration over a range of conditions as wide as possible. The solid Injector a A (x10 4 m2 ) 1 2 3 4 0.757 1.145 1.193 1.069 1.086 Table 4.5: Area and correction factor for the injectors line represents the relation given by equation 4.2 with the values of A and a given by the least square fit. Figure 4-17 shows the difference between the data and the fit for the different test conditions. The other injectors were calibrated at room conditions only. The calibration results are presented in table 4.5. Different orifice sizes were used during the successive tests with the narrow injectors (tests 190 to 196). Table 4.6 summarises the injectors' mass flows and temperatures for tests 191, 192 and 196. The same information for test 181 is given as a reference. For all four tests, the free stream flow temperature is 2000 C and the test gas is air. The mean flow Mach number given in the table is that of the free stream flow at the location of the injector exit nozzle. For test 181, the injector is upstream of the contraction and the Mach number is 0.07. The long injectors reach up into the contraction and thus the Mach number at the nozzle is higher (0.11, given by the streamline curvature calculation presented in section 4.1). The first orifice plate used for tests 191 and 192 did not provide enough flow resistance and the injector flow was not well matched with the free stream. Both the injector Mach number and flow velocity were higher than those of the free stream. Another plate with a smaller orifice was used for test 196. The design goal was to match velocities and not Mach number, in order to decrease shear and reduce mixing as much as possible to get well defined hot streaks. The velocity matching, which depends the speed of sound c = VTRT and thus on flow temperatures, was made assuming a mean flow temperature of 200'C and an injected flow temperature of 40000C. Test number Test gas 181 air 191 air 192 air 196 air Mean flow temperature (oC) 200 200 200 200 Injected gas temperature (0 C) Mean flow pressure (atm) Injector pressure (atm) Mean mass flow (kg/s) Injector mass flow (kg/s) Mean flow Mach number Mean flow velocity (m/s) Injector exit nozzle area (x10 - 4 m) Injector exit Mach number Injector exit velocity(m/s) Mach number ratio (injected/mean) Velocity ratio (injected/mean) Mass flow ratio (injected/mean) 400 400 440 365 3.41 3.40 3.41 3.35 3.82 3.86 4.97 4.48 1.312 1.336 1.332 1.336 0.0578 0.0195 0.0196 0.00996 0.11 0.07 0.11 0.11 30.5 48.0 48.0 48.0 6.59 0.092 46.6 1.31 1.53 0.043 1.27 1.27 1.27 0.166 86.3 1.51 1.80 0.015 0.171 91.8 1.56 1.91 0.015 0.086 44.8 0.78 0.93 0.0076 Table 4.6: Injector flow conditions 4.4 NGV Exit Flow Temperature The NGV exit flow temperature was measured using the downstream total temperature rake. All the tests presented here were made without the rotor and thus the downstream rake was located directly behind the NGV's. The temperature data presented in this section is therefore a representation of the NGV exit plane temperature field. Figures 4-18 through 4-21 show the temperatures given by the rake sensors for the four tests considered (181, 191, 192 and 196). Sensor number 5 was inoperative except for test 181. Figures 4-22 and 4-23 compare data from sensors 4 and 6 of the rake for tests 181 (short injector) and 191 (long injector). The hot streaks produced by the new injectors are narrower than with the previous design, which was the intended outcome. However, the peak temperature is not higher. It is still around 2500C for an injector exit gas temperature of 4000C, which means that there is still significant mixing between the hot streaks and the free stream. Figures 4-24 and 4-25 compare the same data for tests 192 and 196. Both tests were performed with the long injectors but the mass flow through the injector was higher for test 192 (see table 4.6). A lower mass flow with matched injected and free stream flow velocities produces even narrower hot streaks, but the peak temperature is further reduced. One explanation for this additional decrease in temperature is simply that the total energy injected into the system is smaller because of the smaller mass flow of heated gas. 4.5 Conclusion The design for new hoto streak injectors has been presented as well as test data characterising the NGV exit flow temperature. A simple boundary layer simulation has shown that the annulus wall boundary layer does not affect the flow significantly. It might however be useful to confirm this result with experimental boundary layer data. The new injector design was successful in creating narrower hot streaks, but not in getting a higher peak temperature in the NGV exit flow. Apparently, the hot streaks mix with the free stream, even within the short distance between the exit nozzles of the injectors and the NGV's. A possible explanation for this is the disturbance introduced in the flow by the boundary layer developing around the injector's body. The angle of the exit nozzle is quite steep and flow seperation might occur, which would increase the mixing significantly. However, the exit nozzle is in the inlet contraction and a complete study of the aerodynamic interactions between the injector's body, the injected flow and the contraction would be necessary before any further conclusion could be reached. 0O O 0 0O 0 o w 0 - |1 CO o Figure 4-1: Streamline curvature calculation grid. Extended geometry with axial exit flow. BDT inlet .300 .260 .220 .180 .000 .040 .080 GRID .120 .160 9 Jun 94 14:17:39 .300 BDT inlet PRESSURE CONTOUR 4 .260 0 uo 0 .220 .180. .000 .040 .080 X .120 .160 9 Jun 94 14:17:39 Streamline curvature velocity 0.02 0.04 0.06 0.08 axial position 0.1 0.12 0.14 Figure 4-4: inner and outer radius free stream velocity 0.16 bleed leading edge free stream velocity for B.L. calculations axial position (m) x 103 Figure 4-5: detail of bleed leading edge free stream velocity x 10-4 0 BDT inlet B.L. displacement thickness (m) inner radius 0.02 0.04 0.06 0.08 0.1 axial position (m) 0.12 0.14 Figure 4-6: inner radius displacement thickness for low Re 0.16 x 10-4 0 BDT inlet B.L. displacement thickness (m) inner radius 0.02 0.04 0.06 0.08 0.1 0.12 0.14 axial position (m) Figure 4-7: inner radius displacement thickness for high Re 0.16 x 10-4 -0 BDT inlet B.L. momentum thickness (m) inner radius 0.02 0.04 0.06 0.08 0.1 axial position (m) 0.12 0.14 Figure 4-8: inner radius momentum thickness for low Re 0.16 BDT inlet B.L. momentum thickness (m) inner radius 0.02 0.04 0.06 0.08 0.1 0.12 0.14 axial position (m) Figure 4-9: inner radius momentum thickness for high Re 0.16 x 10-4 BDT inlet B.L. displacement thickness (m) outer radius 0.02 0.04 0.06 0.08 0.1 axial position (m) 0.12 0.14 Figure 4-10: outer radius displacement thickness for low Re 0.16 x 10-4 BDT inlet B.L. displacement thickness (m) outer radius 0.02 0.04 0.06 0.08 0.1 axial position (m) 0.12 0.14 Figure 4-11: outer radius displacement thickness for high Re 0.16 BDT inlet B.L. momentum thickness (m) outer radius x 10-4 I.r . ........ . .......... ... . ... . . . . .. . CD ........... . . ..... .. .. ........... . .......... : ;......... :............ : ....... . /.... . ........... // / /: . . . . . ... .. . . ,,,....... . .. ............ I;........... . . . . . . .. .. . ... . .. ... ..... .... ... ....... .. . ... ........ ............................ : . .. .. . .. . .. .. . .. . ... .. ... ... - O. / / CL . / 0 C "0 co - 0.2 / . I 03 " I 0.02 IIII 0.04 0.06 0.08 axial position (n 0.1 0.12 0.14 Figure 4-12: outer radius momentum thickness for low Re 60 0.16 1 x 10-4 BDT inlet B.L. momentum thickness (m) outer radius I 0.02 0.04 0.06 0.08 I 0.1 i 0.12 0.14 axial position (m) Figure 4-13: outer radius momentum thickness for high Re 0.16 Figure 4-14: Long hot streak injector desihn. CtC Figure 4-15: Long injector design. Detail of the nozzle. orifice plate calibration corrected mass flow (mA2) 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 orifice pressure ratio Figure 4-16: Injector mass flow calibration curve. Solid line : theory, o : baseline calibration, x : high pressure, * and + : high temperature. orifice plate calibration deviation from theory (%) 10 8 . + ............ ...... ............... 4 4 o .................................................................... x c:i . X S . .......... ............ -2............................... 0 0 ......... ............ ............ ........ ................................... ................... -- -8 -10 1 1.1 1.2 1.3 1.4 1.5 orifice plate pressure ratio 1.6 1.7 1.8 Figure 4-17: Injector mass flow calibration, deviation from theory. o : baseline calibration, x : high pressure, * and + : high temperature. 65 turb181 spot #3 downstream rake 100 ' 30 35 40 I 45 ANGLE (DEGREE) 50 I 55 60 Figure 4-18: Test 181 NGV exit total temperature survey. Downstream rake sensors 1 through 8. NGV spacing is 100. turbl91 spot #2 downstream rake 260 240 ,220 C 200 L! 180 CL 2 1601 140 140 ... .. .. .. .. ... .... I ....... ........ I .....-.................... ....... ....... ......... ........... .... .. .... . .... .. ... ...-........... 120 10,,rrl 15 ANGLE (DEGREE) Figure 4-19: Test 191 NGV exit total temperature survey. Downstream rake sensors 1 through 8 (except 5). 67 turb192 spot #4 downstream rake - 220 0 0 200 180 qL. M 160 I- 60 65 70 75 ANGLE (DEGREE) 80 85 90 Figure 4-20: Test 192 NGV exit total temperature survey. Downstream rake sensors 1 through 8 (except 5). turb196 spot #4 downstream rake 260 240 . . .. . 3 -220 C, O 200 180 ............. ... ............. ......... a. 2160 LI 140 .-.. .. . .. .. . . . .. . . .*. . 120 -- 1rlrr ANGLE (DEGREE) Figure 4-21: Test 196 NGV exit total temperature survey. Downstream rake sensors 1 through 8 (except 5). TT5 04T 180 30 35 40 solid-turbl81 45 angle (degres) dash-turbl91 I 50 55 60 Figure 4-22: Comparison of NGV exit temperature at 49% span for large and small injectors. TT5 06T solidturbl81 dashturbl91 30 35 40 45 angle (degrees) 50 55 60 Figure 4-23: Comparison of NGV exit temperature at 14% span for large and small injectors. TT5_04T solidturbl92 dash:turbl96 270 260 250 240 t oJ CD CD A :. 230 0 -. 0220 E CD 210 AII .- .A I . ... ... .. .. .. . ... . ... .. ... ... .. ... .. ... .. 200 190 . .. ... . . .. ... .. ... . .. .. .. . .. ... ... . . . . .. ... .. .. .. 180 angle (degres) Figure 4-24: Comparison of NGV exit temperature at 49% span for small injectors with different mass flows. TT5_06T solidturbl92 dash.-turb96 270 - 260 250 ............. ... 240 .................................................................. 230 .- . . .. . . .. . .. . . .. . .. . . .. . . C- ES220 () 210 ..... r I\ ........ S I:.. .......... II I ... ."" ,.............. :t ... ':' __ 2001-. I' 190 ............. -.-..... .. ........ ...................... i \I i . r - " l i 180 L 0 angle (degres) Figure 4-25: Comparison of NGV exit temperature at 14% span for small injectors with different mass flows. 73 Chapter 5 Particle Image Velocimetry 5.1 Introduction Particle Image Velocimetry (PIV) is an optical diagnostic method which consists in measuring velocity by recording double exposure images of a particle-seeded flow. A high power, short pulse light source (typically a pulsed laser) is used to produce double pulses and illuminate the flow with a light sheet (fig 5-1). The recorded image contains double exposures of each particle passing through the illuminated area. The velocity is given for each particle pair by v = Sx/St where Sx is the distance between the two particle images and St is the pulse time separation. This technique gives a 2D picture of a whole area of the flow in one shot, with information about the flow velocity as well as angle. The early research done about PIV dealt more with developments of the experimental technique rather than with fluid dynamics, studying relatively simple and low velocity flows. Seeding techniques, image recording and data reduction algo- rithms were explored and refined over the years. State of the art PIV techniques as of today include high accuracy velocity measurements using advanced data reduction algorithms to locate the particle pairs very precisely, and 3D velocity measurements using stereo imaging techniques [15] [16]. The technique has now matured and the attention has shifted toward application of PIV to flows of engineering interest. The goal of the investigation presented in the following chapters is to obtain high accuracy (typically 1 to 2%) PIV measurements of the flow in the rotor passage of the Blowdown Turbine. The experimental set-up and its major components are described in detail in the remainder of this chapter. Chapter 6 discusses the data reduction and calibration methods, and the preliminary results. 5.2 PIV on the Blowdown Turbine The four essential elements of a PIV system are the seeding, the light source, the imaging system and the data processing algorithm to sort out the particle pairs. The first three elements will be described in detail in the next sections. 5.2.1 Seeding Seeding Particles The PIV technique explicitly measures the velocity of the seeding particles which are imaged. For this to be an accurate measurement of the flow velocity, the particle mass to aerodynamic drag ratio must be small enough so that the particle acceleration matches that of the fluid to the desired accuracy. Transonic flows, with shocks and strong vortices have high accelerations, which makes this problem even more acute. The particle size required for PIV in a transonic flow is about an order of magnitude smaller than for an incompressible flow. A typical particle size for transonic PIV will be less than 1pm. Figure 5-2 plots the velocity profile versus particle size for a flow through a M=1.5 oblique shock wave inclined at 9.25 degrees [20]. For accurate velocity measurements, a particle size of the order of 0.5pm or smaller is required. While aerodynamic considerations demand that the particles be as small as possible, the intensity of light scattering decreases sharply with the particle diameter. This makes the imaging process much more difficult for smaller particles. The scattering is a function of the ratio of the lihgt's wavelength to the particle diameter, polarisation, scattering angles and index of refraction [19] [18]. In the present case, the angle is 90 Size (microns) 0.65 0.50 0.35 0.35 0.50 Composition Latex Latex Ash Latex Styrene Source AFAPL MIT Gonesh#2 incense Duke Warwick University Max temp (OC) 100 100 > 200 100 100 Table 5.1: Summary of particle seeds degrees and the wavelength to diameter ratio A/d is the most important parameter. The scattering power or effective cross section is plotted versus particle diameter on figure 5-3 for a wavelength of 0.5pm [21]. The strong dependence of scattering power on particle diameter can be seen clearly in this plot. The choice of seeding particles is a trade-off between the two competing requirements exposed above. Several options have been explored and are summarised in table 5.1. Another criterion in the choice of seeding is the maximum allowable temperature. The eventual goal will be to use PIV to analyse the impact of temperature distortions on the flow field. For that purpose, the flow will be heated using the temperature distortion generators described in chapter 2. The seeding particles must be able to withstand the high temperatures generated by the flow heaters. However, this is not an immediate requirement since the first tests will be done at room temperature. Two types of seeding have been selected : styrene which has a particle size of 0.5pm and a maximum temperature of 1000C and Gonesh which is 0.35pm in diameter and can withstand high temperatures. Styrene is large enough so that the imaging is facilitated, and still small and light enough to provide accurate velocity measurement. It represents the ideal compromise and will be used whenever possible, that is to say when the flow temperature does not exceed 1000C. Gonesh incense smoke particles are smaller and more difficult to see but has to be used for hot flow measurements. Seeding Procedure The seeding particles are introduced in the supply tank before the blowdown test. The tank is filled up to 1 atm and then seeded with the particles. Once this is completed, the filling of the tank resumes and proceeds as normal until the desired pressure is reached. The paticle concentration in the tank is monitored during the seeding using a Laser particle counter. This device measures the size of the particles contained in a sample flow and counts them. The readout indicates the particle repartition with respect to size. This allows to monitor the seeding concentration and to check that the particle sizes are uniform. The styrene particles are originally in suspension in water. They are injected in the supply tank using a TSI model 9306 six-jet atomiser which pulverises the water solution and releases the styrene particles into the air. This process injects some water into the supply tank along with the styrene. This can become a problem because of the decrease in temperature that occurs during the blowdown. Assuming that the gas in the supply tank is initially at room temperature (250C) and with an NGV exit Mach number of approximately 1.3, the temperature at the NGV exit is -500C. At this temperature, excessive moisture will turn into ice particles which are much bigger and brighter than the seeding particles and will saturate the imaging system. This problem was actually encountered during the first tests and solved by heating up the supply tank to aproximately 5000C before the test. 5.2.2 Light Source As can be seen on figure 5-3, for sub-micron particles the light scattering efficiency is about 104 times smaller than for the typical 10 to 1001im particles used for low speed PIV. Therefore, the same factor must be applied to the net optical throughput of the system (light intensity, image recorder efficiency and sensitivity). For this reason a high power light source, typically a laser, is necessary for successful PIV measurements in a transonic flow. A Spectra Physics GCR Series Nd-YAG double pulsed laser specially designed for PIV applications has been chosen as a light source. It actually consists of two separate laser oscillators which are combined electronically through a digital timing system. The pulse rate of the laser is 15 Hz, which means that 15 PIV images can be taken per second. This laser operates with a Q-switch system which allows very high pulse power outputs : The Q-switch comprises a polariser, a quarter wave plate and a Pockels cell. Applying high voltage to the Pockels cell changes its polarisation retardation characteristics. With no voltage applied, the Pockels cell does not affect the light passing through it and the quarter wave plate acting in conjunction with the polariser introduces a high loss in the laser cavity (Q-switch closed). Applying a high voltage to the cell modifies its polarising characteristics and changes the Q-switch from high to low loss (open). During Q-switch operation, the flash lamp used to pump the laser excites the neodymium (Nd) ions for about 200ps with the Q-switch closed, building up a large population inversion. At the point of maximum population inversion, a fast applied high voltage opens the Q-switch. The resulting laser pulse has a width of less than 10 ns and a peak optical power of several megawatts. The global energy output per pulse is of the order of 100 mJ. Because of its very short pulse width and high power output, this type of laser is very well suited for PIV applications. The transition of Nd ions produces a coherent radiation of 1064 nm wavelength, which is infra-red. The laser beam is then passed through a single harmonic generation (SHG) doubling crystal which changes the IR beam to a visible 532 nm wavelength green light. The disadvantage of the double oscillator configuration is the high accuracy needed in aligning the two laser beams to prevent two separate light sheets from being created. Spectra Physics has overcome this problem by performing the beam alignment within the laser head prior to the frequency doubling operation. The main advantage of the double oscillator approach is the flexibility and control it provides : * The individual power density to each oscillator can be adjusted separately. * The pulse power output is not dependent on the pulse separation time. The timing of the laser is controlled with a Stanford Research model DG535 digital delay/pulse generator which controls both the time delay between the flash lamp and the Q-switch for each laser, and the separation between the two laser pulses. This time separation is critical for data processing and must be set very accurately in order to get good velocity data from the PIV measurements. Figure 5-4 shows the two pulses sent from the delay generator to each of the laser oscillators. The pulse separation was set to 0.6ps. The error from the generator is less than 1 ns, which translates into a precision of better than 0.2%. Figure 5-5 shows the signal received from a light sensitive diode exposed to the double pulse from the laser. The oscillations are due to the response of the diode and do not reflect the laser light intensity. However, it is clear from this figure too that the pulse separation is exactly 0.6ps. 5.2.3 Optical System The optical system of the experimental set-up can be decomposed into three major elements : the light path which brings the laser beam from the laser head to the test section of the rig, the sheet-forming optics which turn the cylindrical beam into a light sheet, and the viewing window through which the images of the flow are recorded. Figure 5-6 represents the position of the viewing window, illumination periscope and camera relative to the tunnel test section and flow direction. Eventually, for PIV measurements with the rotor in place, a second periscope and transparent NGV blades will be used to provide a complete illumination of the rotor passage, as shown in figure 5-7. Light Path The use of fibre optics to route the beam from the laser head to the test section has been investigated but proved impractical because of the high power output of the laser. The fibre could not withstand the instantaneous pulse power density of the Q-switched laser [17]. A rigid direct optical system has therefore been chosen. A set of dielectric mirrors is used to reflect the beam from the head output into the sheet-forming optical system (periscope) as shown in figure 5-8. Periscope The design requirements for the sheet forming optical system are manyfold : * Small size. To minimise the flow disturbance, the optical set-up should have a diameter as small as possible. * A sharp light sheet it required. The success of the experiment is strongly dependent on this. * Pressure seal. The optical probe must ensure a suitable pressure seal for both vacuum and high pressure conditions inside the tunnel. Figure 5-9 shows the design of the light probe. A diameter of 6 mm has been chosen for the end optical components to minimise the effect of the probe on the flow. The probe has been designed to mimic the operation of a typical telephoto zoom lens. By using two cylindrical lenses operating together, it is possible to form a light sheet and adjust its focusing position to be exactly in the plane of interest. The two cylindrical lenses are on separate mounts for that purpose. Rotating the whole optical assembly changes the orientation of the sheet with respect to the vertical direction but not its focusing point. This adjustment is used to ensure that the sheet is parallel to the plane of the camera. Finally, a 450 dielectric mirror turns the sheet in the perpendicular direction. The pressure seal is formed at the probe exit by a 1 mm thick 5 by 6 mm a/r coated window, glued in place with epoxy resine to ensure proper sealing. The position of the periscope and light sheet relative to the tunnel is represented in figures 5-10 and 5-11. Viewing window The viewing window is positionned directly on the outer wall of the flowpath. Therefore it has a cylindrical shape and its inner diameter has to match the outer diameter of the annulus which is 55.12 cm (21.7 in). The outer diameter in then determined accordingly to minimise the optical distortion of the window. A ray tracing program was used to calculate the optimal outer curvature of the window, which is 78 cm (30.71 in) [22]. The window assembly is composed of four elements : * The window itself which is made of BK7 glass ground and polished as shown in figure 5-12. * A stainless steel frame to which the window is glued using epoxy resin to provide a good pressure seal. See figure 5-13. * An intermediary aluminium frame on which the window is mounted. See figure 5-14. * The main frame, also made of aluminium, which is mounted on the test section of the rig. See figures 5-15 through 5-17. The periscope and the window are both mounted on this frame. The intermediary frame holding the window can be disassembled to provide easy access to the inside of the test section for adjustments of the optics without taking the main frame off. 5.2.4 Imaging system Two imaging systems have been tested during the preliminary phase of the project : * A film camera with Kodak T-Max 3200 film. * A video camera connected to a PC with a video board. The film offers better resolution than the video (4000 x 4000 pixels instead of 780 x 500) but is less sensitive and takes time to develop and digitise for data processing. The video system yields useable images instantaneously. During the set-up phase of the experiment, a short turn-around time is needed in order to make the necessary adjustments to the optical system, seeding concentration, etc. Therefore the decision was made to use a video system for the first tests and move to film to get a higher resolution if needed once the system is tuned and operating properly. Pixel resolution Cell size Sensing area Dynamic range Scanning Clock Pixel clock Horizontal frequency Vertical frequency S/N ratio Minimum illumination 768 x 492 11.0 (H) x 13.0 (V) tm 8.8 x 6.6 mm 67 dB 525 lines, 2:1 interlace 28.6363 MHz 14.31818 MHz 15.743 kHz 59.92 Hz 50 dB min 0.5 lux (f=1.4) Table 5.2: Pulnix TM745 CCD camera specifications. A pulnix TM 745 CCD camera is used with a Data Translation DT3955 video frame grabber board mounted on a Dell P590 Pentium IBM PC compatible computer. The Characteristics of the camera are given in table 5.2. The frame grabber board is synchronised with the rest of the system by a negative TTL signal on the parallel port of the PC. 5.2.5 Rig Movement The movement of the rig during the blowdown run was a major concern during the initial development phase of the project. The positionning and alignment of the optical components are critical for a successful experiment, and could be disturbed by an excessive movement of the tunnel during the run. Video images of the facility were taken during a run to estimate this movement. Optical displacement detectors have also been used to confirm the video measurements [22]. The movement of the tunnel is essentially in the axial flow direction, and has an amplitude of approximately 1 mm. The transverse movements were found to be small enough to be neglected. The light path mirrors and sheet forming periscope were designed to accomodate for this 1 mm movement. The light receiving end of the periscope is wider and the lens has a large aperture, so that the light sheet will not be affected if the beam is not well centered with the probe. 5.2.6 System Synchronisation This section descibes how the tunnel and its 300 ms useful test time, the 15 Hz pulsed laser, the video camera, and the frame grabber board are synchronised with each other. The timing requirements are : * The laser needs approximately 2 s (30 pulses) to warm up to full power. * The flow in the tunnel takes 200 ms to establish after the valve opens. The useful test time is during the following 300 ms. * The video camera operates in interlaced mode, which means that the frame grabber has to store the two fields immediately following the laser pulse in order to get a fully exposed image. * The whole system must be set in motion by a single triggering signal coming from the BDT timing system. The following scheme has been adopted : The BDT timer which is controlled by the main initial trigger is used to send two triggering signals sl and 82 separated by a 2 s interval. sl is sent to a H.P. 33120A function generator which outputs a 45-pulse 15Hz negative TTL burst. This burst is sent into the Stanford Research timing system which drives the laser. 82 is used to open the valve and start the blowdown. The 2 s interval between the two signals is used to warm up the laser. The video camera operates in interlaced mode. The CCD array is being exposed continuously but the frames are transmitted in an interlaced way : even lines first, then odd lines, and so on. At 30 frames per second, the camera sends a field (a half frame with only the odd or even lines) every 1/60th of a second (16.7 ms). Each field is exposed during 1/30th of a second and sent during the next 1/60th. Figure 5-18 shows how this works and how the laser is synchronised with the camera and frame grabber. To get a full image of the particles, the video system must record the two fields exposed when the laser pulse occured. The frame grabber board is programmed to acquire the image as soon as it gets the trigger signal, and to start with the next available field, regardless of wether it is the odd or the even one. The board can differentiate the two and reconstruct the image accordingly. The frame grabbing trigger signal is sent from the Stanford Research pulse generator which drives the laser. There is a 1 ms lag between the grab signal and the laser pulse. This ensures that both fields acquired by the video system are exposed with the particle images. However, this method does not work if the laser pulse occurs during one of the transfer gates of the camera. To prevent this, an asynchronous reset signal is sent to the camera by the H.P. function generator 5 ms before it starts the 15 Hz burst which drives the whole system (fig 5-19). This introduces a lag between the laser pulse and the transfer gates which remains the same because the camera frequency (60 Hz) is a multiple of the laser pulse frequency (15 Hz). This safety margin ensures that the laser pulses occur between the transfer gate pulses. Figure 5-20 shows an overview of the system layout with all the components and the connections between them. Seeded Flow HLight Source 0000 0B 0 OO Oo 0 EO ,0 L-- Beam Forming Optics 0D Imaged Area Light Sheet Camera 440 420 400 380 360 340 320 0 2 4 6 8 10 12 NO MAL ODISTANCE FROM SHOCK FRONT (mm) 14 16 Figure 5-2: Particle response to an oblique shock wave, for different sizes. 10-10 o 10-12 Scattering Cross-Section (m2) Nd:YAG (532 nm) S--- Rayleigh Scattering ' - Geometrical Cross-Section 10 -16 II fi i i i i i i . . . .i I 0.5 1.0 Diameter of Sphere (Micrometers) I I 5.0 I " " " 10.0 Figure 5-3: Scatter cross section of spherical particle as a function of diameter and wavelength. 86 93 Acqs Tek Stop: 5.OOGS/sr ET *r Ch 1-Ch2 Oly 600.26ns . . CAIi a . . . - . aie . . .- i V i ."-.. . ,U Ol .i .ni i ., l 4 i 9 V 19 Apr 1995 17:49:10 Figure 5-4: Seperation between the two Q-switch triggers from the laser sync box. The interval is set to 600 ns. TeK Run: 5.00GS/s ET Sample r "r- Trig? L J 20 Apr 1995 11:02:52 Figure 5-5: Seperation between the two laser pulses, detected by a light sensitive diode. 88 / Figure 5-6: PIV optical setup. T. Rotor Nozzles S>Casing Flow Upstream Periscope Window Figure 5-7: Setup for PIV with the rotor, using two periscopes and transparent vanes. Test Section Window Camera Periscope 87 Mirror Optical table Laser beam 19.5" 19.5" 1-4 7 1 9D 24" 610mm '° LASER rL , 78" 1981 mm a a 32.25" 819 mm U K ! | Roor. Floor r= ii 495 mm ' *'- 4- -- Focusing Lens Sheet forming cylindrical lenses Steering mirror and vacuum seal window Figure 5-9: Illumination periscope with sheet forming optics. Frame Window Laser beam Laser sheet Figure 5-10: Relative position of probe, laser sheet and test section. View 1. Laser sheet Perscope - - Viewing window Figure 5-11: Relative position of probe, laser sheet and test section. View 2. E i 4.330 i : 0.25 inch radius 2 : Inside curvature : 21.700 +- 0.O inch diameter 3 : Outside curvature : 61.42 - 0.01 inch diameter Figure 5-12: Viewing window. ----- r - Figure 5-13: Window frame. 95 [mI Figure 5-14: Window holder. Figure 5-15: Main test section assembly frame. View 1. 97 Figure 5-16: Main test section assembly frame. View 2. 98 7.25 16X .500 MIN R10.850 4.995 Figure 5-17: Main test section assembly frame. View 3. 1/60 sec Laser pulse and frame grabbing signal 1/30 sec Even Field A Field A is transmitted during the next 1/60 sec Transfer gate pulse Field B Field B is transmitted during the next 1/60 sec Figure 5-18: Synchronisation of laser pulse and video camera to obtain non interlaced images. 100 Tek Stop: Single Seq 10.OkS/s IT Chi High 4.00 V Ch1 LOW 80mV -- ' ' ;- ... ...-- ....... ..- - --+"'' Ch2 Freq 15.01500 Hz Chl-#Ch2 Dly 5.20ms Low signal amplitude Ci 2dd 2 2.00 V MU s .4U V 18 Apr 1995 17:26:06 Figure 5-19: Lag between the camera asynchronous reset signal and the first pulse of the TTL burst. The lag (chl to ch2) is 5.2 ms 101 Trgger from BDT da aoqulon sysem Q-Swith 2 Lamp 2 Lamp 1 Frame grabbing as-nl Video BNC connection Figure 5-20: Component synchronisation layout. 102 Chapter 6 PIV Results and Data Analysis This chapter presents the first PIV images obtained from NGV only blowdown tests, and the data reduction algorithms used to process them. The data reduction software and calibration method are presented in the first section, the PIV results are exposed in the second. Data Reduction and Calibration 6.1 6.1.1 The AutoPIV Program The Windows AutoPIV software package (APWin) has been created at Warwick University, UK. It processes digital particle field images to sort out the particle pairs and calculate the flow velocity and angle. The particle data taken at transonic speeds tends to be sparse which makes the processing of individual particles far more attractive than the global processing usually applied to low speed flow PIV. There are several advantages of working with a direct image as opposed to the more conventional 2D Fourier analysis approach : * The particle image field can be very quickly reduced to a sparse array of data points. Thus, instead of large digital images of 8 to 20 Mbytes the data is reduced to a single 30 to 100 kbyte vector representation of the remaining particle field. 103 * With such a substantial data reduction it is possible to apply a very computationally intensive, and thus more accurate, processing to evaluate the information. * The data are of a sparse nature. If a global Fourier processing method is applied then a large area of the image must be sampled to acquire sufficient particle data. In sampling a whole area as opposed to individual particles, localised small scale flow features such as shocks or wakes can easily be averaged out. Background Noise Elimination Each particle is processed in turn. The first step in the algorithm is to discriminate between background noise and particle information. This is done by removing all the points that have only one pixel of information associated with them. The next stage is to sweep a small radius around the remaining particles. This inner radius is set by the user and defines the lowest velocity which can be found by AP. If another particle is found within this radius then it is assumed that the information is erroneous and is the result of light scattered from the surroundings. The user can define the lower and upper velocity limits in which AP can search for. However, in practice this band is kept as wide as possible to avoid pre-filtering the velocity data and missing real flow-field information. Particle Pairing AutoPIV allows the user to enter the parameters of the experiment : flow speed range, flow angle range and particle size range. Within these parameter bands which are kept as wide as possible to minimise the risk of data conditioning, AP inspects the particle vector file for potential pairs. Effectively, two radii are drawn around each particle : an inner radius for the velocity minimum and an outer radius for the maximum. The existence of a second particle within this domain yields a velocity vector. The presence of a third or more particles in the area invalidates the pair. AP is also capable of making several sweeps through 104 the data. In each of these sweeps the particle pairs which clearly satisfy the choice criteria are saved and removed from the data field, making the identification of further less obvious pairs possible in the next iterations. Particle Center Estimation The initial step of the data processing is to find the particles and reduce the image data to a vector containing their coordinates. For that purpose is it necessary to locate the particle centers, and the accuracy of the final velocity measurement depends directly on how precisely this can be done. The image data is a gray scale image. More precisely, it is an array of integers with values between 0 and 255, 0 being black and 255 white. Several methods can be used to locate the center of a particle image [15] : Highest Intensity Pixel This approach consists in finding the pixel with the highest intensity value. It corresponds to the maximum of the 2D light function and the center of the particle if no distortion or large noise is present in the data. No previous operation performed on the data is necessary, and the algorithm is very simple but the error is as high as 1 pixel in each direction, which gives an overall error of 2.82 pixels for the distance between two particles. Center of the Bounding Box This method involves a thresholding step and a black and white edge detection operation before the construction of the bounding box. Then the center of the particle is simply the center of the bounding box, which is the intersection of the diagonals. In this case the error is 0.5 pixels in each direction, which gives an overall error of 1.41 pixels. An illustration of this method and a comparison with the previous method are given in figure 6-1 for the most disadvantageous position of the particle pair. Center of Mass Considering the light intensity distribution of an image, E(x, y), in every pixel of coordinates (xi, yi) the value of the intensity is E(a~, y;). The algorithm finds the center (Xz,yo), also called the first order moment by weighing every point by its intensity value : 105 (X = 0 1,YO) = (soyo) j(xi, y1 )E(xj, yj) Ej j E(z , yj) (6.1) (6.1) This methods needs, as a first step, to identify the limit of the area over which the computation is performed. Since this process will be strongly influenced by noise, a pre-filtering may be necessary. Bidimensionnal Gaussian Fit Assuming a Gaussian geometrical image for a spherical particle and a constant itensity of the illuminating beam, the intensity distribution of the focused image will be well approximated by a symetric bidimensional Gaussian function (see fig 6-2) : E(x, y) = Ae-(('-o)2 +(Y-yo))/(20,) + k, (6.2) where A is the amplitude, o is the variance of the radial symetric Gaussian and (zo, yo) are the coordinates of the center. k, is the noise pedestal. Starting from these theoretical premises and using initial guesses for these parameters, the algorithm uses a Nelder-Meade bidimensional fitting method for the particle data. The process is iterative. The inital values of the parameters do not affect the final result, but have an influence on the convergence speed. This technique has several advantages, the first one being an accuracy of 1/10th of a pixel in the position of the particle center. The Gaussian fit method can also solve images distorted by noise, saturated images, overlapping or incomplete shapes, or even half shapes and interlaced frames. An aspect ratio different from unity (which happens when the pixels are not square) in the image can also be corrected without loss in resolution. Figure 6-3 shows some of the types of particle images that can be solved. However, in order to work properly and yield the desired accuracy, this method requires the particles images to be at least 5 x 5 pixels in size. 106 6.1.2 Image Magnification The AP program can locate pairs and calculate the distance between particles in pixels, but to convert this information into real velocity data, we need to know the resolution of the image, in pixels/mm. This is done by taking a picture of a ruler with the same imaging system used for the experiment, set at the same magnification and through the viewing window to get the same optical distortion as during the test. As can be seen in table 5.2, the CCD cells are rectangular, and thus the image will not have an aspect ratio of unity. To account for this, pictures of the ruler in both the vertical and the horizontal directions were taken. Figure 6-4 shows a line extracted from the horizontal ruler picture. The peaks are the graduated lines on the ruler. These are 1 mm apart. The resolution in pixels/mm can be calculated from these data using a Fourier transform, as shown in figure 6-5. The first peak of the Fourier transform is at the frequency of the graduated lines on the original image : f = (n - 1)f,/N (6.3) where f, is the sampling frequency, n the index of the peak on the FFT and N the number of samples. In the present case, this leads to the number of pixels between the peaks (period) : N m n-1 (6.4) The accuracy of this method is limited by how precisely the peaks can be located on the Fourier transform. The low sampling frequency and N number on the spectrum reduce the accuracy. Using a higher harmonic of the frequency gives a better precision. for the ith harmonic, the result will be : m= iN iN n-i (6.5) whith n being i times greater. The result is the same of course, but the precision is higher. In the given example, the first peak in the FFT is at n=21. Assuming 107 this number can be determined wih a precision of 1, this leads to an overall accuracy of 5%. Using the 3 rd harmonic gives n = 83, with the same absolute precision of 1 index, and a relative accuracy of 1.25%. 6.2 Results This section presents the PIV images obtained during two blowdown tests (204 and 205). Two images were taken during the 300 ms useful test time of each run. They are shown on figures 6-6 through 6-9. The particle pairs can be clearly identified with a bare eye. 6.2.1 Timing Figures 6-14 and 6-15 show the NGV exit Mach number as a function of time for the two considered runs. Figures 6-16 and 6-17 show the laser firing signals and the frame grabber signals which indicate that an image was taken. For each test, the two images that are analysed are number 3 and 4. They were taken during the period when the NGV exit Mach number was fairly constant. The repeatability between the two tests is good, both in terms of Mach number value and timing of the laser pulses and image acquisitions. In other words, the images for tests 204 and 205 were taken at the same time during the run, and in the same aerodynamic conditions. 6.2.2 Particle image size Figures 6-18 and 6-19 show a 3D mesh plot and a contour plot of a typical particle image from test 204. The particle image is approximately 2 x 3 pixels which is too small for the Gaussian fit processing. To try and overcome this problem, the camera was moved closer to the test section for test 205 to increase the magnification. Particles images for this test are shown in figures 6-20 and 6-21. The particles are larger (4 x 3 pixels instead of 3 x 2) but still 108 test 204 205 Horizontal resolution (pix/mm) 37.1 55.7 31.0 44.8 Vertical resolution (pix/mm) Table 6.1: Image resolution for tests 204 and 205. too small for the Gaussian fit algorithm, which does not converge on them. 6.2.3 Magnification The resolution in pixels/mm, which indicates the magnification of the image, was calculated using the method exposed in section 6.1.2 in both the horizontal and vertical direction, for each test. The results are summarised in table 6.1. As explained earlier, the particle images taken during test 205 are larger, but a high price was paid to obtain this result. The field of the camera was greatly reduced and is too small to cover the whole region of interest in the flow, and these images will not provide very useful information about the flow field. The images in test 205 cover only an area of 14 by 11 mm, instead of 21 by 16 mm for test 204, which was already slightly too small. This shows the limitations of the video imaging system. The film has a greater resolution i and will provide both particle images of the correct size and a wide enough field of investigation. It will have to be used eventually to get accurate velocity data. 6.2.4 Seeding Density The seeding density in the supply tank before the blowdown was 4500 particles/cc for both tests. This provides a good data density on the images. However, there is a lower particle density region in the images, which is probably the wake from the NGV blades. The data in this region is too sparse to get useful flow field information. Thus the seeding density might need to be increased, but this might make the particle pairing in the rest of the image more difficult because of the too high particle density. 14000 x 4000 pixels after digitisation of the 35 mm film. 109 Test Resolution (dpi) Aspect Ratio 204 941 0.84 205 1415 0.80 Velocity range 200 - 400 m/s 140 - 1700 Flow angle range Table 6.2: AP image scale settings and parameter ranges for tests 204 and 205 Test Image Mean Velocity (m/s) Mean flow angle (degrees) Number of iterations Number of pairs Mean Particle area (sq. pix) 204 205 3 4 3 4 341 334 361 364 74.0 74.2 72.7 73.0 4 7 2 3 250 285 92 172 3 3 4 5 Table 6.3: AP results 6.2.5 Preliminary Data Analysis The APWin software was run of the images with a bounding box method to locate the particle centres. Table 6.2 shows the velocity, flow angle and particle size ranges used in AP. Table 6.3 shows the number of pairs, iterations and mean flow variable results for each image. Figures 6-10 through 6-13 show the velocity vector field generated by AP. The data density on these plots reflects the particle image density on the raw images, with little data in the wake region. The accuracy of the velocity measurements is not better than 20% because of the low resolution of the video system and the method used to locate the particle centers. As can be seen on figure 6-22, for test 204 the mean distance between particles is approximately 7 to 8 pixels, and the error is 1.4 pixels (using the bounding box method). The images of test 205 have a higher magnification, and thus a better accuracy than those of test 204. Figure 6-23 shows that the distace between particles is approximately 11 pixels. This explains the discrepancy in velocity between the two otherwise identical tests. The mean flow angles agree well with the design value of 74'. The magnification ratio being higher for test 205, there are less pairs on the images 110 and the particle area is larger than for test 204. The particle sizes indicated by AP are smaller than the ones found in section 6.2.2. This area is determined by AP by thresholding of the light intensity function, which captures only the high intensity pixels instead of the whole particle image. This however emphasises the need for larger particle images and thus the use of a film camera. 6.3 Conclusion Chapter 5 and 6 have discussed the implementation of the PIV technique on the Blowdown Turbine test rig. The technical challenges and the solutions adopted to meet them have been listed. The first results have been described and analysed. The images obtained so far do not allow velocity measurements with a satisfactory accuracy, and yield a too low data density which makes the investigation of flow structures like wakes and shocks difficult. However, although these images are not expoitable for flow field analysis, they do show that it is possible to do PIV on a transonic transient test rig like the Blowdown Turbine. The fact that images were obtained indicates that the optical system behaves as expected, accommodates for the rig movement during the run, and that the synchronisation of the different components is correct. The system is now operational and the next step will be to move to a film camera which will provide better quality, higher resolution images and flow field velocity and angle data of the desired accuracy. 111 _ .: :. _ dl:. dL distanc between -article centres us=ng highest pixel inteIsUty d2 - distance between ;ezticle centrzes LuingZ cL c .the bCund ng..box. usrngC~p-~~t 0 tt~ebou.....o Figure 6-1: Comparison of highest intensity pixel and bounding box methods for finding the particle center. Normalized intensity 0 2 z (m -* Figure 6-2: Gaussian approximation of the light intensity function of the particle image. 112 I1P~wl d) a) 52 cA,0 a *a 7-MO aY I x- l 200 Sa Figure 6-3: Examples of particle images that can be processed by the Gaussian fit method : noisy profile (a), saturated profile (b), incomplete shape (c), overlapped images (d), interlaced image (e) or distorted image (f). 113 Light intensity of a ruler picture (gray scale image) 18 0 ... >,160 C180 : ...... .. .'" . ........... ....... E 140 - 120 100........ ...... 80........ . 60-............... 60 40 0 ............................. 100 200 300 400 Pixels 500 600 700 Figure 6-4: Horizontal line from a ruler picture. 114 800 C9, 0 ... - .... ... -- ......... ...... ..... .......... .... ,. ..... . ..... ....... . . ... . . . •"::: -. :..... :,,lr' :.:.'.:: ':. ,: .!.. ':::", ,:,, ... 00 :::!i : :::: ...... . . . .... .:, ::: ' ': ":' " " :':: ': :'::: . . !~ ::::::: : : u~ . : '::: : ; ' : ' ":. o A ... ... . O F :;:::.:: '': ''i ",::. ".' iii i ii?i iiiii :::: :i:::: :~ .... iiii i ~ iiiiii iii !i~iiii~~ ... " ii !ii! 0 " ::: ~ .................................................. o ~ iiii! ii~ liiiil j O 0) ..................... ............... ." U-.. ....... C) 0 ~ "'''i l -.................................... :. FY: ::: ::::::::::::::: ::::::::::::::::: *............................. r bO) ,::,:.-.:..,,,.. ; ,:: ....., :::: : : - :::::::': :; 0 0 ~%U ......... ............... ,: :::::: :::::: : : ::::::: :: :: ::: :: :::: :::: :: :: . ......... %.'., .. ,m I. 4-. r LOW0 '+4 0- ,. 00 C O c ma (00 t um CUO~ 1 b 00 FbO V ...-.... . . -- " - . ' ". . .:-- s ; g .. C"-"' " ---- 'a '--. Z41-iA " ,.-. S " ... . : - :." .. . - . . :.. ;. " '- -?---' ... " _.... - .GL.;' - & "' -.-. . . .. ......I.'.., "' ri . . 'S.''-= . ...... .: ..... . V4- #.. .......... ......... . T_ rr - A.. i ... ... .- ...... "'"" " :':i" "' :;" .... '- - .~ .- . v. *..- Y.. Is ..... . Z- N - - -. ...... : :';--. °; oa.-'. .- . . ru..e, *:. - . - -. .- :::-:...' -..- . *";~'l:.: ~- -- c-----' -- -' ..-. . . - .- -'. . .. r - -I:.=. '. -i_, .. ..-.:.. .. .- :-;-'. .. ... . .2 , .. '.a-:. "' ..-. .. .. ... -- ' ..... - ...... ": :..:".. . . "" 116 . .. , -. ":'-: •" ..... .r . ... . .. ,-........ . ... .... S' .' , _' . . . " _.. ..,.. . ", ,. ; '-°.' -. . -- .. ..,-, - "... ::. . :.".. . .-/-.".' "- ; .. .. ...... .. . .a -. :... . . -.. .' ---- . . .-. r-. ._-' a'. " .. .:... ;-, T . a . :i... ";, -; -. -. ...... : --.:'" '_' ',.-:; . . ' '':7 ;: ;---.I!:'.:.....i.. : -.,a, ... .... .-'--: ,a .-,:...-, _ =. ,..t .-..-.. . .... .=a --. ' :; g • -.:.•-,. .z* :r , =,:.. li,.._.. , o . .. , • - ,...- . ,. : .-. . ' ..,. ,..- : • {..:..,'t'-;:; - . ... " .. .--. :,r~ . :-;. •-,. i - .... ... i= .'= . :-.. ,=:.'l = :~~ii~.ii.~- ...-":.:;-:,.,..: -;-.-.- '-, :-'> .-- .. ... -- L-T .. -am ...... PV imge,- est 4 imge-M -3 =: Z.-~ -'-~~~i........ =:."1C~3~~ Cr , -._,... j ,.. .-... .... .... ,, ... ...... ''~ ' "'": c:; l,':. '"; "":... ''': .- - -*.... . ..a . . Fiag~re -6: :--.!" "..: :[';z= .. .. : .:. -c ' ..... c.. ...-i ... :.:.:i...:. 1.~:: l ! ag A :'V: , -;,...t'- Fare " .? . ... .... .;, : . • ./ , .. .:'... .: . . . ......•. . .: .a- ,, . .:.. ,;:. ,,.,m,.- -:•~ 7,:.--..,., 6 6 . - .. - :" ..-'-.... . .:-: -. • ..-.-- Raw PYimagetest94ma _X-- .... ...... . . .... - - , _ ,!n.,r_:!M.. ., - r- :,.tm.=.:r,-..!.i==, ffnr! - 4.-** :-777- -!7 -.-. "--,-v* ... ... .:V:--nr .........:. . =---7::: '- , .!?E-7-*!' - ----*.: .f:n:-.-:.--.-!4:-.,.-.Ii-.-.---':-:. .-:",?Hft;:::-. :.:^. ;.......... .... .. ... _ ..:.-,.--. .... .., :. ..... , --..:! :-..,,;: __ - -.- :--77--7,7!---L... ... . .-. 6'..'. ., ": ,-.-.,:.,..;.:;^ *,..-..%t. .._...._......;;;;*,*!%.-n...i::=:.,-:%4lI,.c7.-.. . ... ., ..-.. .;... -'. -* " .0-.. , .-:, !..!.,-.; .:-,. _ 1* ...,Z::. "---*-r * ,!:ii--. ,:--.:,--.; "',n-a_ - . . ::,:t:=.;::M.:1. ...- ...-.!r.., ,... v;.. , '?_...Z:l.;,". -.!.%....-I . .. . - .. *.---Z..,.,.-.-. .. " , * -.-.-.-I.. . : % . .. . -J. .--. %___.. _ !, .. .. . . .. . .... , , .. .. " :.C... . _ ._ _d.p4.;-.._.... ... O., i-7. _.,.. * '!::$ -=:::xl- . .:-_ -"-*.--:ii*----!?!!'7h::"*: -?', . . ..... . .--... :.,.7. - . . 7_zWml:-ii-- -::. ,:.: *- ..-. ;..P.' - , -7, - -4. .4. ,, .. , , .... !:. . . _ .. - ...4;;."... . " . .. .,.:"*. .. .!(.- ...*-..: . . !V.. m. '.. - - . ,-- " .,.... :- .. %.*. ... .... ::.;.:.:4.,-Ir.:,&Ijt:2 .. ,-.. --': -:= :.M,=7".., ,.:.. =n:- .,, -J6..F.:r: :: : !";-.-jo-.::_ . .- -..... - .-- - -: ----! .. a:::.--=Cmm. ..:. .. , - ....-'... _.;-.. ; , -L,-_ - .; -.. - --:-, .,-,- "...--_.....*P4 .iAo.: : - "' :- ..:_ -- "- * - r. :........ ...:-- . - :. .. .1.,-. .11.6.. .. ::,. - _. .1;.. . . . -.-...-. ... "n -_ .. ... . -- I .. . ... ,::: I .-:.-.4,.:-;.:. * .. . -:0..t l:, --;.-ii-:. ,;.b., ._ :1 - ".j^.:-.::..j.l.,. ., * .1. .;W .mz.; .. ... ""-;r -S ?,. d '. . :;:;t - -- _ - :-.-,-T-. .. :-.-lV, "."... Atl*o".,.%!..;...,.-,..:*.:"r...- :U-.*X-.. ... 7..-. , .. : _,..-.: W . .. , . . ,::; .. . . . .-.'*..w.:, :..,A--:...., -.::.:-:5il '. -_ - ... --- ,.,,,, ,.".,:& 'i;!*,:*:..*::" ... t _, ,r y..... . I.... .4 - . -. -.:., . ., -!e.. ,::,M:WM.... :1, Ise . *..." -,:.* ,'.=Z.. _f., -. "!.- ..'.... ... . .*._**.:.-. -. --.:-...:. .. '... .....-,. ___ . ..:! 4 .:.:-' , .'.-.-:.:m---,r .:.:, ..... r! -M .. . _ -.-. ._ -- ,:... -. -:,.i:-,;%.W-:.;..::. ,---; -. . Tr=_ .... .... .. 0,-A.t.----, - *1 ... ,-. .... MW -m"-;!: ..A . x. :.F!:,?,:.. I!....:: tmc.:. *... ..---,... , ...- ,. - . ;IV .....---..;=. . ="_r.-... ., . - .- . .- . .- . . 6. ..- ll.-L:."-ni_ .. TW, .. 1.%:_ ;Z's -_= . ..... _.-.:A-. 1. * _. . i 11 7.,:=:::.Sr ..n .;i7, .:,.,.Wlr --..--mr -. nm i --. -_,,,,,."...-..l ... I.. -4. ,. . : - ..-*..-:-.-: - -, - ..nm A A -J ...R * .... 4. ... ... .. .;:..,n7e. ...-. .. _... . ;-;.Z.Kr.ft...*;...n M.?*..... ..--..;; .-:::. . .1a:..=._ , =__,.,: ,..:T, --,'-' . -..-....:. ".41. . .: .r.::,:n-:.::Zww,;*.4;-! 4. - ... "'.4 'd -"....f ' d-A .. .. ,. -zi; . ?!.2v-.,%.LM"2" -ij.'. _;".";iij,Z.v , : *'*'*.!,.*;-.:.*-.'-:--. .:-, - - ".,IV -*!.:..=. ...- di,.. ----.. .-.. ., .. W.iE .. .m_ . '! .Sv. , _. V.. S.. ... 4; ,X,Tid .. ., 'A -: _.. ... :j.... . 111, &C.- . . , * a ::,i . .. . .. NVA" .%j l: . .. ..W a ..-- *, b!" i -_U-. .,;.-: ,. J -Js- I - .. - ...I.. - -.."_.". . ".. _L_;.,.,i ".. . .. ,=:'l, -;W -- w-.m.....i.44W pi -,-... ... :,m m .. "!j . . :*..... ZWZ . .. mu-,.:7 :" -----x:. _. , ...... .... V-1 -- - . , r... .-. :....:---.-.- . ... .,:",., .- ";.- - ;em. -Wl-.,# . ,'-.. ': . . . ... .. ..:. . ;;P,*...'. :. , .x.."K..P"?,t.." P.0jig' ,t6 ,. "I-. ...: Z-': _ + ..... .- .j:0;;?:j!jjl!_ .4 ... . ..... --. 4 " 11 -0 -M iy W -!!,.,;.:.O,.;.A----. - .-.., mWia .. - :::--- I--- , " . , :!**:':-cCf*::.:Y , , ;.. .... .. .... . M I T'I* .:. - -4 .- .,...:::cwvx*.=.:Zr-. .. .- ..... " .'... :_L:7 .--.:-, .= .. ... Z. - *-%Lz&.. ;7 ." :- '..... . .: .. :. ...- ..- .. . ..,.-: , .. ,.. -.-.. . -:44" .;.='--_-_ . . . ....; - .- - .. . . -I ..,:.* - - - . . --.:=.;..-. :.,: , ._.-,:-! 1-_ - ; - . .". .. ,"_ '.-- .- :: ::. .0.. . -. . .;_: -,-;:" .. . . ::.. .;.Z..'.... .,- - -- ". .... ,:. "'. i_= 1- '.;.- '. .... .... ,_-.. ..;w..-, ;. .."w;;::;;. ..::.z-:...:::F-.:-..:.-:--:, -.- -T .. r- - - - -- - .. . _;'... 'e" -, . ....- !.::=.-.i. ., - ., _.- .. .:, ...-. - Mnu:.. ,. ! M. -.,:-,.j-c*"* __jL. . --4-W .:K.:C:: -.&RW Z%4v.*Z ..j...... _-._e,- .m"'..;.... I.." - ___ ..-Kc .-I. ...... _ .-. :-, - M=: r-!er ,_, - - -- ,,.;ir ..-'-n.- '. ...-... - .. .- . *-:...M.4.;L.m:...._ - .-. *...... ,".. ._' . . _.. l _!M.:.-;;:.,.--. .2-ml .. * , - W..j . -, - . Al)T '. ;. I !: .. =--- .g='kW..::.!r. _r WrlWl . M., 4Wf,-.,:,:, -.. .",! Ib '-.... ";i i ...... .. . . .;. -Z:.:z _ . *,._.,::m .. .. _; -- :",;:,?!, *: :--_:4;, ., .. :-:* . 1:..;, ;.-..: .. -% . ..... -, ... .. , . ,.. "xg" tm"Jq%:--... ... .. ...: . . . . : ... : ., ..j .. .Rj,*P.iS=., .1 _....*"*:,.*--.--. .. - -i.. - - - ... " - -,.7 . .:7... '-'. .. lk4,,j= ... '. . ; . . .....1-VIX: -, ; a. A .. mxi..Z. .1-1 ;-.2 . ^"i;k,:*:,YI . ..... ..-4 I . .. ..- -*-::.----.TH:..r _.. .. -,,; -r --. - ..:.:-- ....*... ,.,. I.4,L,?h;rv.Pm=*bm4-;4-40- *u%-.*" . .1. . .. ... ,-". 7:.. . -_ . _..- -....--- - .. .. . .. . - . . 1. .. ; -- ,.- ;,, :.-...-7: t-.. . . . .... - -, ,RCWH-4: _--z._. .. ... ..1 Wc k-l .- I-'W --I-... .. .e--.,--..- V m iv ,W.%;Lle . , . 4 . 0=-?4*i,; i-----! . 4.. . .".. .. "!...... *.* . . . .. . _... ;__ q- *.. ... "_ . ,-, z-;: := .. ,... *i%--'*C.ln.... .I...... .... _r- .;' ... .. =..,.;.. ... - ... ..j,;,"." -- .-i. .! .1. ._- . --W MM --, -> S'.*.--W,-.=Or-..:." - .4 . ... ;. '. -.. 44. -t-... ..I . R"t ..S C,... ..-.,4:72 M; %. . .. 7;-'.11 , . . .:--::.".j'4.*:7z'Azi: -e W.. i;d-"kv. _ j..::. ,': ..' -_ -._.... . tx l-k'.;U-n , :. ....;X _- .. ;. #Z%=.- m=o i.,- --T,.-.7-. -.1. ... ..,. .... , . .. .. _.: :..". .-!-:ii;m - . . ._.. . - ::- :?:,J. , -. -,; ..Z.-" i.4': ,., .Xx2ouj 11 _l -.0, - -*-. - L_ .._X.l.:. .. . . J. -Ii . ,5 . 6 1,'.Z - Kc .. !7 ,'i;v - :M.,!vrW . .. : _ YA. . .. . ,-. . ;. ..' W.U4$iF .. ... ..., =r.c. .:::, , - .,::.:: X", * -... - ;. , O.I.* ,I.-,*,&,. Lid"S,-=I'*' . *!. -.7 ..... "". : . .:r . '". - I .r= tv;F.m.'., ::iiiiro.v . = -"*- ,..ZWA3-R Iw; -- ': .- ... rL, ;::.,.l.:,--i---" .. ... , pip". .:'. -.. . , **.. ..-. .. .. .&ij ,.. .'- ".-I Vb --.----,-- -:-:L:ijiE: :-M:E: -.;-. . --.:-,-. . - !9-%;A.V.---,._ - '- ..-- !;.,*_-'7 ., , ; ..r::, i.N., ;Ll-i-. ... ka.,:M-ZWk$.kr,'*Y--. . . I . -:;% .;,.::::,.:FF -t%,. . "' . jp , - : .. .. .., - -;7. . . . .r.. jTi* , . . . 'f= W_ - - e., , .. ... ,, .... I _- ,_..&;4.. - ..--9 . 1 " . . ". - " ___ . ........ .. :-:., '.%,., :-.:S,..:: . . - .... .4:..... .. . - -: . . .....Z.... - , l ... . -.. .__ . .i .:."*'.- .':*. -:, ;!--. T.. .,. - 4:EW-..'.--. ;.! ...t I. * ...... . ...... , --- ;___..__. . 4.. ._ ," __ K. .if" 7.,-7 -*-:7" . . .. _; 14.-=,-. .:.rt.. ... .: _ , .4 ..--.;-;;-:iiv:.. 'pe i H ,:,.:f,...--%..'., , .. "I'.44 . .g-_.gtr4Wm . I . -=-,V$i*Q _:". .!: " 7" o--l.... ... _. .! ;q.,= * f ! ..:... '... ' .:.-....-L, -. ...... . ;..J.. , - -" -;3 .....-- :: -... . . . , . '.w.... " .. _ _... :-..-.-!F ...... . ;:::*,., : --. Z --. - - -,_ -_---.-, -:.:V-;,;.W:P-#M_. -0 .N. . ;_"4zu_* - ,:,;"-.;-.:.:.::-%.l . r . '**ox.*V:! , , 3p.*-- -A., . .7. . . . ........ ..... __ . . #.,-. *_ .,,;, .. -.;:.:...*Tr-! -:'.!'.:=.-.. -r-T I . .. . , 2"tl" .. .261a....=. -- P -.",:il"*-A--,A%:T-.. ^ * . . ... %.I.* . -=Ae-__ .. . - . I., " : loall.: _ - .. - --.-..-.-., ;.- ;::,.-=e:-. '.'.--.. -, ...... '. .:. -. . "t*r... '-! _... !'"...... .. . . *.. : . l.:.l.* . .6 -.Ww'.... .', 7. .._ ..... i;,i.;AA;__rW., - ; . 6:4!:-"O . -=. . zx.-; . , .... j" C* . :% .. .-.. . . . ,..,:'" -%::= . . .,.!. I-= :.. - ;..... . . . . .*r . . **. cWe-W--. 17::!*r*t'-l-x'. - I *.X I ... ... ... :T....- ... - . _. . -:.,:...:r-4 . , P, . . ... . - .. , ,:,..: : : a . ; ? .. ' . ., - . ... - b,:_.;. .... .- --. :C: , . .. ., -.tn:tnh -I,--_7.7= .. I . :- - -. . : .. ... - .- * 1 .. - __ .. .. " . -* - . ..; 3141. - .. , ..C*rk:,-;.; ... t -.. !-.5, '. *'.. ; -.. !..-.!MklA _-i6_;i.i .I , 1), .. -.. U . il .-:t: -.c m%4, i;*4 - A..." P-J., kiluo-4 .. .... .. ... . 1. .;i . ... ;;L. . .. ..-- .. .. , =APNO-N.-JecM, .1 3 .. I I &I . i - ; ;il ...... *? = b _. ..;,je_;... , ik. -_ . -.- z .,.-:: ,.,:..rL :.4:._-..;It: - ..... . -. - . j:, *.;A- -*- :_;...;-_: : .. - % :.--. " .... . M.-..*.-a--:;-,. .. .. j , ",..: .;A -- -- -----.. .j=- -!;:,,, jja M , .* .. "'..1. .. .. _.. .,.. ...... . ,. . . .Wn. . . - , . - - -L... . . e., . . I* .. . .iii.. %'.:.-,:-.. ! .. . . .:. - N;- --.& l ..... .. . '-. -.. -. -..Ir ;..";*Z;;vZ..i;,..e.!T ....... ::-10, . . ,. . .. 4:z, , .. ... .. .... .. _k- *- ,;, -,- , - .::.,.- 7'. --,1'_ ..1:-.. , __ f.? _.; , -. *2r . . ii. I. : -.ob . .. ... iFAouz_--.-. . .l*.q,,'-l6 . .. . . " , ,-*, ** -1 _. .1-4 M,.,.4W,:. I .. . . ... .,-,:.:. . ',"-'--''.,-"" =.;.t'T.'-jr'.a'r-"-., . 7'"I" --:-' :F--=:'*:,;--,.-::--.:u=.W. . . . . .1 . .... 7 . j..c .,..;;A,..... .:t$._,_i! ... --iT-F,__. .._ .v ......... . .. . . . .X9 ;,;.v . . . :. _ -- :IL::-..-::l:. .. .:.7:-:,,: .;ivc -,-:L-.;pY-.% -4!." 1, % . , - * I:.--,--. *'-.. .... -.:"A,;k -. ". '.- - . .1.. -0*41,C , 'm "!::i: L4r -A - e-Z '$!".' , ..4Rv.,P.. -.-v-__ -i--..i7,::, .,_Z !77 - --. ... --!-.-....- ....-".---r jud.HtLl ...-.* _=;. _. _. Z% ... ...... .. '.. .. .. ... . --:,c.;Iz. ... , , ...0:; . , .. ... i --M .. -: -1. ,.-. .-; . .. .t .;.,l .. ..-.;,,f -!'M .. %!; 49 ...... -. % %:-_-.;,;-V..,,:2r_-.: - -.-: ! - !. I-- - .- -':, r - ..... 1, 4., 1 Ift . W ., ;n*u.; -& vv-77.,_:: _i;;K_.-;;f .. --! . ,XN : .:::* - .A ........ ... -.=:W.::, . ...:: :..-. . -^:--_A -; k-j:.:...-.::.-;:74!:; 14-!;.:_. ... .... ' %!' . . I,--:.,::'... ,;...4,Q -,. _in -n..". -W... _.. -- "Mr, .,, __-A ... .. -*F= !-. . . A. I ..( -.r4iM77 .' 1:., . ..*- ."._.. - .: -. . Ky. ... .- -'.' .. . SA,_.... . . .-*-.:L:," . .. :.:; ::. -W. ..- - .;.-. -.b. ... .1_...... , .W. .: ..- - - .. , - . - .. . -" - --. :V2'. , -. . --n- r *I... . - :--*-:X,;ZN-r-p'l......ro ..- .,., .1. - -...... gg._,'t.r: 41 A.. -It..."_-.Q";. ;.:.; - .;K.:V..m .... -- .. := . .,.: -.. .;. .1. ...XT,-V=, . .. _ .='!y' ... *.*_.; -&R"!.&0i=..::L .:.. %7 -7'-- .-. . .- ... ... .. Z_ . ..... . .-.. ;., . _P" ... ... . , . . z '.. ... -!.. '.*- I- _*W,".-llft,_f,.r^:,-. . :jk:; ,I;v -, Y-, --.. H"I.I.*-...j rl._ . .:. 17 .:= ;m ,-; .. .. , Z;k ,:.v_._ _ .-. ,.;.T _:- : ..-. ,.:: j'..-:C.,:.-t-_;,H. .- '.'.- -. .;_!"--. :,. - . .. .. -.! -.- i 1. rv ... . - . .. .. ;::. __- .. ,;. ., S, . . ... ...W1 ,5---l*.;,W:vc;, 'f-n:" . .. .. . _W %X" -.. ., ......... :i; 9 : . 't . .f;u 1 7'."! _ . -.-. -. :.%. --. -..%"*.;;;t .. "C'... .11.. ,!" -V-..,...I.':.zj - -._ -. :--X... , !,a. .-..;;X.%*"Kz,6-i.,!.". .,- a .,-. ... .: . .- .4.1, , .. .:L.-; C;,4 .- -,.;..,Z;;".=-.----O,-:,%,?.iia...:.;Z '110.74-m% ." r .. , 1... I.. I :;. '_:. , .z 5 . .... ,----- - --!,A, 17_.2_1 -. . .. . , :. ;%_ - * , . '-" " . . *,%.CX." . . .P ,I . . _ . .... 2.....-. :.:. . . !-= . .... ... , , . . % i Z" -M.--5 ; -, .... . 'N A, .;. . - .. ...... n.,. , Z",.,:.*C_ l:-:l;:x". .... :- .: .. :. . :.. . . ... ... ; -. ; -' .----. ..-w---4 .2. ,"lik .,.;;-.-, - - *-- --. _,: :*; Z j%-71%1,-'...; -:_'. .=.,. .,. P.1 4: 1 , A,--K K .- - . .. . . -: . .m . ... . -=,=:!:%_.i L"j,--i _. . - , . :1.. ,;. ... :,:7, .;;N. ., . ... - , , . I.:: :.,, .,-6;;L_,YQ':'%g-.-:;..; . .. .. ... . ....... ....- " ".. . ...... '.v%.x. .... . . . .. ... .. . ... . .- L lzrs 1.* -= ;-t.., q-l; --M.f. -.i; -.. .. ..._i-.%p.lZ...'....-.;rP-4 -,W7 '" .---. :7 .J..., : , -a.- .. ***- .... "r . . 'Ir- . ,-.., ".-- ,u-4l-*xJ%-.. - :-- . . . , ... , ...: .. . . -,-.="-;: I-!..W .. St.w .X.*, ..!;Mn . _- ... ' -:! . , _ -. -. '* . . - -- ... . "'141; " .. . ......W ., -, , . ., in , ..: Mr-C."%% .. - -!W .., - .... '.. . -:.hg.J.., %!.... ..'. _;Uxl; . C.zz. " , -.1.'j..*..M ::X ** I- liar'-:.. , r!.%:.; .-.;-:. ... :1 . . .,. _._ ... -': . .. .. r, -.. K- ..::. . . ..... !r:-i -4. .t io. ..!, ::,,,,Z.5;;rv., zg*, . _..... --.i.--. - ...... -,. .. ; .:%. -_..... - ._-.r . I ,,,,.qv..C;O:r . . ..... .- . -. . :-. .. . . W..;- :,::.-.*,V . , .7;. .... ,4.;6,_.-W z -A.;i . .?TT.;;:. .....! ,.:::-. ; .:-..%. ::. . .-.. . 7. : , .:- . ! .J... ..-..."., ... Z:4; :-: %EQ::.a= . . !iRm.n .. .'.I?f6i .. ........ . : .:. A...... ! -a. ....-'V---- -- _.. "'. - ,.-;;:-.r'l -%-..-r -....- -! .. .._! . _... .-:*jQ * *".;,...:::-:!&. ,,:-:.* . . ... . " - :. " - ._W.It)-%, , !*'2e; -'Ai;.I% . . . *- '. .r .;Ag.% ___x U ::,. .,.-_:, . .... .... ,.:;U!.:, !% ".,; . : :. %-.. .... ":T. "tfo'. - -. A. :!-7 .... 4.".X...f. ...... .. ...: . ..j!*..i7,;z,.: . ..r. :, t,-.:r. . - --.Z..,A"j----. . . . . -........ .. . o. . 4.. .... -n-T.. . --o . .... , .-... W_? ,.m..fgr ::,_,..._' - - -. - ': I %.. . :...i:,--! .,.. I -ol: *.,',;,:-::.I. . .1;1 .. ... L. ..?&,. .:.* , "'. -,.. -. ....:r-... '__Jfj,.;;;;. I ..*.: :l . .., ..... ... , . .. . ...1 . . ... . .. . * . I... .1 ...:.: '-o ...... I :-" ;:.. .:*,:::**: .',-;4 ,&..:",v-,;".;;, .1 ...o..,k--.'_-.iWAv:X.:,., .. .. o.:-:.&,:..-. --.--. l.,'.",n.9!lN - - -: .:i-'r'--. = .I..... ---., ..:. -".' _. . . .. . - m:,,-* -. " _* .- .. :;. 71 , o C- ... , b.. .,§ ...... . . ' - ... ......,o,%.,*%.' . " J" -7.. o.".! - -. oI =**' . , :Lkr, .;. . o -. .o '.... ;.I; - ; .. ... .. 7,.. . .. .o.. .:.. , .... ., -.:. . . . ,--":: _;W." .. 6-o:%;kS. ::. -. ;ijjj: ._: -,..... _= L ,_ :*%iove= -. ;r., ,4 't a,.:i ,jx; .. .. - .. . .. ,. , ... . . i.m. . .. :;;.';'-,..:.ii;,.,-!-..-.;-.:K" .:-.. : . .... W -. 3 ..... ::.-',---:-.:--, *. ..... * _. .. ....... "" . ... 1 ,..*.lv.. :., *-'. -. :.,.---..Ij', ::vi.. !....:..7 :3..!** ,,,.!.. ! -- ,-1, . . . .- - _.1 :*.- '.-. I . . .1 ,-.-: :."-r . . . _*. _. -::.,:'.........2,Ill;j ". : , 4gv. I.- . . - Ze *. :.... .9- :., :je.:.. . .OV-,, 1:3 - . t=j 4. 'r '5G -- . -;j - -% .jv",*. --! - - .rq%_ - ,- . *.?% W".7z% .__ 1-.$, . . ,* . * -.-. -*-,.... C--- z4kk'-.. ,-..--!..-r-:f.:zW.1 &.' ... . . .. .- II - .. *. .*7, . --.. '. I .. ., " _-, - T:-" --- . .. - ... .. . . . . - -.-- .--.--.= .. --. -=- ux;. .;.A.,--',.;j,:".r_ .. ' .6,.-;.=-. %Ila:" .'-. " . T 't., : , * "-,'*,. -: .- :., . . . .;,*.:7: . - .... . .. : - - Z..--L*- " P... ;M __ ... _,.I-'i .,- , - *..-. * , _.:r:. .. ...-'L..;--7..;;::,. . -: - :7-, -. i;R % :vow_ .7.,.,:*..--,M ,.:**: - - .,P,* . . . , , :.:*. " ? " .;- %. - . ,Z------. -:-,'-.-,*..-*, . . ,_- ;. ".--.,- .;W..... "'.:". , -, ,. , I.N :.,---..-.;,,4* . . ..:,!..., ... ..: , ...... .-I ... - ?., ..X;l,.,Z. ..__.;,..:. .- it--. .. !V-..: ..-- !__ % . -. . . .. . .....7. :.t *: . .* ! . .!*.,:. . .. . . .. . : +..... yor ':' -- ... 77'. - -.--, -- .. .:. _. . .. ... .. . .... i-.,:. :....-.'..: .Nsi.. .,."... .. W. .;:..*--. -1-- J-.Al. Alt .H;::;-4::!:.:'. . ... . ....:L.- ...-;*;.;.IW--%m-irlit'%_ .I...... I -. .Z*42i 2 ti,. ---: f: ! -m- - . .-, -_: . -- - "._;. . .. -r.. ,.: ..... . -- .... .!.;:.. . . - .- . 4 .." :.- .... ,- i;.. U. .! .. . .;A:: ..... , ..*.. ... ... ... ........ ... ..k ..,--. *11.*. x.:.:0 _ ..e v. %. 3.7 ....... .. ..** .X ;.. "- ... A.... .- . ... I::...- qi . ..'..;_ . __ III- . . .. .. $ - -W .- ;:r-r4jft.l "K -.r. .. - %L.-. W . ;;"I,. lj!yl.,.r - L.-ka, -;C-!t-_' . 2. * , A*., ._ * . . ---.:.!- .. _ -%-. _ . -=*-?- ..'--_ *,.;--VAj =;:-:? ; .,aw-'_-Owi . .. . . ,-.'.'--..*Zlm' .-. ;- ... - ..;.....;Z.. - -;.; .,.::l. ::-:_'. . . -:. .-. ,.-.. * -...-- --_, " .- V%"I!r:L , ., _..e--.-.,"N.. ,;.*:- .-zl ;:e,;.,O.:A :: .: _ - . . .. '14. -i.:=-Tt:.-.:..Z. - --- ::-:! . . .: f. ...%%. t. . .. 1:'.... :*...i-.:--.;, :::.:,:*,--, ,--.z;i:-, . .;.. - . .'.6 ,-.-..._ -.. _. 4;. ",-!,-*:*--:.:.-.*.. -v-=-!. , ..,.. .. %* I* ; -. . .. ,-.%-.::i.;"; u--i.. ...-..... .... , -. ..- ...;-:.:.;:i-.1_ _... .. :om-;:-.;;,:;. ....-.. .. . :: , .2 , 1..; M * ;::., -N--:-.N?.-.,lK7*6.., -;_..... . . -..._:-=v Zt. , ;j6-;7,_l;-_;.,::--. . .. -..... . . , . . '*..;.....;F:-! .. ' . :I .. 3& :* *a,; ... *.,. ..1.. .., .-'.:, =..-.-:-I= -! -Z!:-::.::.,;:Y'-;-_*--* . I'.-::.. :' --74 .. , '. --;A .; =?. -.... . __.,._ -,*..-:-r '-"'-- - .. ;iKA.A,--*-un;i -. . .:::' ,.:w . f.... . .. ..... ._ :. . . :t ... :. .N. ... -- , -I. ..--.;...:.;:. . .:. :. .::..-.,,7:'=-:_--..-_:!%' * ". .. I .- .. . . .:.;"... . . ,:. .%. .. .,_-iP...-A,_ _::. Z. ,4: .. ;:.,=, ___ . T?;n:f!..;_.;*.*" , lzr_,_ .S.:* :,...:o .% . . . _ -. . i . _. - : ,A.-:-_ - .;-,;.:..! ,W. '"' - . 1: .! , . ,:... z,tW .;.ti-Ap *=;.A _-% ;,e_!&_ .,;;M p_"*l!V. '.. - I... J-11 --,:-!--- :- .. w-n-, -. . .rl .:...1 . lnl .. _ ... e.nir-1. .I..-** - - K- ,f-:I* :i.;v -. ;;" - ::.....,, - .0,;-*.-._,fX - .MllKl_ :,:,.i,'..!: .,X, .,,.,::=ZW -. e..:. --.. !--. . -I.-:.=:. - -, ... - _... ", . . ,4.'.. .:....._-;Z-:: .,.Z: -%-..:: = . .. e.:. .,.. :....... -m . .. % ':a,.z::.; -1:7-:_,:. _ K_ ::-. .--. :: 11- ; ... . ,.-ii%; .,.4L4ti;--:Itt-,5, l... I;.--,P,-*--=;;.:. . i .;, 4;." ;sw,,;,.e.-H..-. " " ...... " .. , .----..... W:;;--.....,..:"*.'..;*.*,! i x _= A --I_-:- : .. C. :., .*.:0, .1 .... - . i:,: -... V_.. I....! ,n' W-'. : 1,t -..... .- _.. .. .. ,:. _;,:-:L*., i--. '- '. !-..v!:lll:*'. ", .-.'.:lM;Q.'O.!%= "",F ,,,. , , i ..::* .; .i . ........ ... .." . .. . L. .A ....:. -- ..;.;.::..-. t .. *!:..:.,.tlnfiF " : -; ,-...... . ::,L-.:.T ___7 IM ... i4xi. , ...-*i.. 'Z:.*...;-..!;.;i--t,-. .. .. .iz - m-.A-_...; .A . I-. : ... ..... ii:O r . .-'4..;.:Zz .V ..: *.:::-:= ::-.-_...j:._;__ _ .....-.- :.;.--!%. QfPcUft".. *N :... -..- --..--- 4 -w., ..b....a, , ..... .:*;P,.,---, Ao=:6.. ":!:t",.-,-T.*--.. ... . :,*. . _UW_:*iW .S:A. , ;:.,. .,. .... . . xll-_i ,-.--;;.::. *1 ....."--"-.....x.:-;!% - -k*m, . -Z;=Zmv.; L=:m .7tmax%,l ' r -G.*- - - ,-o. .. 0 --. &-'. - O'.... ,..I--, :&-_ Qz: . . . I.jo=NNriM . . . ........ ;; N_ :.;.Z.,!,.:j;z 4*;;.,.-;:;_:: . j.4 2A ., '? ',.,-m'_ , --4 %-.- _ -,; t.:" ... . =: , _'...... - .;-- !ft--' d ._.q - :.;.n . , -;L.%J*i .i, , .W . f M -6.* . - --'g. .:: !:,, .. )-.j ,.- . -:-.-. -.W , " . &-. .%Z-rc*;.Q .' - !. -JW .. tir;! ",W. * * '. -!'.=:.: Z44-=-_"..; _'.xz"?.. ONY -:F; .: ! ,'r....- .A&_M.Tl! _To;v 6.v ....;;Q .I'... :-; ?:", . Ift . I% ,.-,.- '!'-W'-.'*' -4 ::--I.... - .-_ - j_' ii, _" _ _10. '. n% .. .. . _z :%;:P.j-?,2x _. A lf_ '-N -..:-..-.- ..- , ;! _.-.... ;44i --"4;:W.4 ^-ft-, x4m *= ..Z,::. * -,+".:i, ..-:.. ... '.. . , .. i .zi = .......... 41H -..-.. r Z :C ., :;; , % .. x Z __ : ,:.,-,.;W... f ." .!...'.F.-.... 1; t . . -m '. ,:. .. ... .. _ lgp . r -3, It - a .Z ;-, --..c. ....-A::,.-1 t -;..-j;;;; . :.V. -; - t::.-=:.4-'j;, - 'u"i'm...;- ,..,Z7.;.6..,,.._._. :,., :'% - .. ::; .v .--... ;._ -... . .... ,_%i--!;X-. ,.. , V. ;:.i:-n-wm%!:* . ... ;:-.. -: *!7:".= .1 :..;... .,. ..-N, .. ,*,:::::k ..-_.IZ4-, , " ..7 _..- .V4.-, - - - - - '=Al N ....:,Fj __ -.*_Ll 7.:, t _ .: . .i _. ..% - . ,.:., -:--:r-:::.---.,t,!.ji-::-,7.7,::,-...;.*, -, -' :-- - .;,., .. ;;.. .:= 6,Np_-i.:: n,, .;......... Ir*. . . ... .. . , aw :.. . _- z -. 6k.-Z ... ; W, i*:::". .. I1-1-S=4..!. .,_ _.*4 -:==-;iz7-- ."bAy .. i= .Z. . ... .%'.-r.. ,:::O-;; . a. ";;Z;!?. :-;, :!.,..":.*-'-..:.% ,....;:, -. ,*, .:., .!.'-R*V.t..Wz.A...... .;:::....... . 7, .;--.* 4..-W-e .. . .. ": Oi::....y":v:,--* .,!...v .......; :t:;;4. n;.*:*",;,*. 61, I*,"-, .,-r '.'41- L: ;-; .... C. --' YXIMA.. .- Kil ... -Uzw ... Y4.,Z-.f_,%i,; I.. .:*'- .. -...-' .::;.. .,"V. . -. - x;; - . ,:.,: .. .. %.f: * ;..1 -:;. .. ."," -'... fT::?!::-.:-.q,! ._ -. .,.., .. ... :--- ;;,- _Z.. .. ". ... ;.. -. * . . ... -..;...;:.. .. "...-, .1.... .. ... , .. i. ..... - '%**-, -_ -_-. ,..;i. -;:-*..:==gfacp.,; ii., , ;;Z,-,! - -. - .--- Q: ",;,. . r, ." -Z ,;..,;;, - :: ... - ."i=7' wpt ;N .. . . V.M .- . " '!? ...... MHM'a-;6W-__*;; "m . t __ -;: * ...... ....... .. . .-: , " . AN '..". "' ... n"-:A *.11, . '. _*_No:;2_ W5 ___-.41 ... . .,.,...--..... :_... . , .Q .. 'I .qL_,. _V::-j: . : .. .. i::..; ...-'i =-:A& .1 !-%P.!!... ,.?s..g-l. ; -: .... .. -.- ,- Mz.;,; v , . E M , :. ,-, -.- '.4e-,. ;;Fv-:.:,'.tj? .. _";....... .... .: .. " 7:...: : . ., ----..C;;. *,..' W- Imr2 _=t_ . -.x.A%;. ;-.K'x'Q ....... .. ...... _. =i&:-:2 ;.Frr2vp= :::,*.'-.,.;,.: :-: .: .. :. l:__!:.!.'-..:'_-', .. .*-,...,l'.":,:::,_Z,... . .. !- .I......... . .. I . ... .M.::; ..'* *1 . . ..;; .t r . S= 1111= S. :ww-. IV W 10 M* Im ..Z25-n-, A ..... I.-... -_, *: ..:: , . ... , :- . .: --. -.4 !i' .:=:: ... S,-;.XMlq . I . ,.::: ... :. : .. . !;;. ;i-,1. , ,4'.''14=*" . , .... . . , ; - ,I as .... , , * " * :, .. ,. -- I -7C A**,'* __ PP . . . .4*v*.:. . -.,FA ... *___., V,,.-... -i..--.*..;:.::-!." _ -... R: ., *l".. .-, - - , 4. -'NA ***.*:,S,. .*_:ii;,: __4;#'gl_-:.--Pz .1 I. , , _ 'Xwtm - z. '., ...'... , -. " !r=.,,.-i '. .... .l.-. A.V-_ . , __l__i . ,!?-.. - I .UIN ,4 1 - "W ..*-T-?-!;: :-N.a:;..;,.-: 2n ; = .- ... i . , .izi -- n m. .. i:... -13 . .. V , -!%Tb ..r..M - - ,al.n. . ... MIMS&'* ,-,*7fq,?..;; ::.x .%W. . . - J -- ;=.--,,; ,.,-.z :. ;" oi 1" - . -q..... . Z*!%U. -1 .. ; , 4pm .: :.= . . . -.:,'..::--!-=_*Yl.YAo .3'. --- -;;,"*a*W:P .. _: .VW ,;.. .-_.-..... ...... : itc . .. -- - - .- . . . .:; V_ .. .- . I',1?1_ _1167vlcl . ..*-. _... _. .:. . ... t a.... -- 7::-,;.-L ---. a A ' - 1. : .---- 1. -.-. .. .. r" .. tS6%jf *:.. -. , ,._.. ., .ThF. . . , p . _ ...- -. . ... : n... _ _. ...-. - . '.... .- ..-.. - - _. ---.. _-..., :: ,--... ; . ... . ... - .;, *_... . ,:-, 4. .W..,:.:D:! *.*_ ,.:- lv.,-7 at ...Ix . .. ' ' . . . ... . %. -:., -, : . lo. . , =-Z.'r %or" . . . . .* . -..-.I- ...'-. , . .. . -e, 'L_' heJ.' .. , . _:_ - .. !,.L . .. ,.; ... , .. :.,.! . -.-.-. _. i . . .-,* ;. . . . ..- . . . -'...._ .....-.-.._ -.. :..- _.-.__ -....... ... _"_UAN .W.orz .. -, ";;.. . . . . :'- !-,TTn..E.r ft. ...I.---. :,-., ;. . .m. .- .... ..... , : -_ .. %. . i . ... .a ..- ': - %. . . J: _k - .. _ -" - ;nr!_.4:z.t: .. ~... :. ,.I.7:'; -.--- , _.o; _Z.. ; ;, -. .% W . ........ '::. ...' ... .. . . ,R n"p rt ; - . - -.. lio; .a- " -' " - . '....'."'.--""...:. - '° - ' ' - ........ .~ :. ...... " ' .. '' : .... '., ....R: ' .,_.--. '. .:.... p '." -= - .. nw - 7.. .""-.. . *r... : -. . ,. ...... ........ . . : _ :...... ..... .,.,t, ..:,.. .. v,,~~~~~~~~~~ . ~~~~~~~~~~~ - ".- •• ..-..-".... ,.• a, .. . . ........ •-.. ...-- :..... :.~,. 1 18. . ~ • > ..o".7 ... ....... ",;;_ - ". . ."..-.o .':'' ' . '.. ',",-. " . " , . .. a=,':' ."-'::: . '.a"--':.. :" ;;' . : -".-" - ". .. . . I. .. " • •.' .=.'. ". .*.:. '"': ..:.;.,:,.1 : , ,m . . .-*. • .r -.. • " , €* -"... .:.. . ... :::• -. .=.' ". " • -'" - ; .. .,... :............ :: ~- ..... . , .. , L;, : . ::~":;~: .... ... I ... ,..' .. ~. ...... ...... .... .... ...... ::, .. ...... . .. 1. .. ": '" 4.* "*",,: .: . . " .. : """ ,:, -;. '; . .:; -. . z: .n .,..... ...:. . ,, : . -. = ::... =: :':": . ! ;: i:-....-. ,. .:= '" -'":.a':."-'.: . ... " ' -; ,:.... ... :" .: :: .:...:~~~ ... .. :+.:.:.-....-. .:._ - ....... :...-.. ", ".-.:' T. :. .. i..g.., : •:_,'. !"!: ,:,'= .-. -:. • "-".. -:.;;.. . -', .-. "."."""';4: " . . :' " " ""• .: , :': "-5' ..i;-a'-a--''-:-'.:::'-',. -... ':i-':2:': ' .'::::: • ...-: . . .• w.. . .., . .a. . . ,*: ... a . . .=p .:..:... :.• s . - • . .. . a . . ..•.. .. . - .. ., . :,.,. .':.: '-.: : :" : ... , .. .. - "..:-:: **,-..,....... :-:"" .. . . |' •:4 ' 0 7r ~ ... XT 7* '..:.r ...•.":"" .::'_.=."-::' '-..:::: :..:.;: ,a=...::: "'r"-;: .. :" . . '.- .- I, ". . .. . .'- .. • .:.:. . ... . . .. . ... -- .. op- : --.. ... ".. '.. ... . •.*..:...... ... '-- ." ...•... . •~~~~ .:. . . . . :;" . . . _ ::: : :' ... - . . .; 'i ..-. :: .. .... ....... ~ ~ ~ ~ .-. ~ ~ .... ~ ~ .:. ~."." '. ...4::': .. -.. ,-.-." .. . ...~ .•-.. ~"'.-.--. .,.:. .~ .: ,-:.,:-.: •,... , .. •., ..... ,;:41;: :.: ;. "." "'.:: •~ 7I ... ~ .*,. .. -. ...:. .. .,.. o' " . . ':': '"" . .' • i'" •. .. . .. .. .. ... : . .:.. • .c .. .. .C .. • -. , . . . . .... a.., .'--- : :i o.. . . " .. .:: :. :." " " -: ..7'."" -: :...+" ~~~~~~~~.... i:..:-:.'2:.::: -. : ... - ' :. ..... - - .. '"" . .::": : ';:%:. -: -: : . ;" ," . ... .... ;7 :: .... .. . a* .. •~4?' "" ..-., .... -. -: .. : .: .. . :., . .'-" . ,.. Figure Raw 6-8: PIV image, test 205 imawe-::43 •,.. • ..... ~. ...:: '-. :.. .-. .. ... ... ... . ,. .::.::. .: ,. . . .•.' " '.... ":. ., :".... .-'' ,.. "... '. .:.-;;::::-:: .: ::: ::.: -""= '::: - . .I.. " ... 4.. . . . .: *' - ..-:--. .. .... .I . - .... .. .: • '" * - . " . .. " .. . .. . ' " " . ' "" . .. . " . ';'". . - .. .. .. ;:" :" • .• ....- ..:--. ,."".....:'. ---: i:... :7.;.-/ -:."•:-7::.. -. ':.;'5 -.- "., "'" '"~. --. - . . : -.' ; • ".. .t ,,!: ": :..i_ ; "=.... -r• ... :.-.. r • . . . :- , .. . .- ' . ... .. . . ...' ' . :.7 .. • :' " "T '".'="1.,-., !'M....-:= . . "'-J:= .... :.. - .. ..- -.-*..-.... .1.A : 4 - . . .... .... t.'.'e. p -- tj it y, ,.:,.N 5' ____." *.-."" . . . : ...,L. .• ... /.. ....~o. S~t4.~ ... t ........... .._.:.,--:,:,-, -w.ww-:... . .::,; ; .:., ,:-:: .,.-reMy.8Al, . ::. SO.. . .r .-., ..... Z*_,L -.. -gal,- -..---,."-- -'t .... ..... ,. N U. ...... , . . . . . . . ,.."---, ..- ,.,,. z..... .. ... . .'. . .. .~. . -- 7 ... . ,.... ." . . ,%d .. . ..- . -aa *. . -1.• - -.- - .u' --' - --,,. 1,'., "..... - .". " "....-:"..... . .,..,..-.,. . ' .:.. . .. ... . . .... ... .... ... . XL...• ... . t-..: ''' ";: .:; ... "... L... ......-.. .. A--' -,: .-... '" ,,'..-- ... .:".. "".. - .. '' ' k"-" '........-.... , . • ........... . .... -,... 44 k.". ... .. " •. . •. •....'".• .... ... ...... .• ...,.-.-.":........................ . .. :''-.'1;6... .. ..... . . . .. r'.. .:S Figure 6-9: Raw PWV image, test 2Q5 image-ra4. 119 b.. .. . C - -- rZ I_ C i I i - -- -. - - '- - Figure 6-10: AP processed PIV image, test 204 image#3. 120 '-. - . C- Figure 6-11: AP processed PIV image, test 204 image# 121 4 . - - Figure 6-12: AP processed PIV image, test 205 image#3. 122 i k I 'I \ I' 1 I' 'g 'I I ~ ' 1~ \ \ I \ \~ I I N' N N N 1 \ \\ I \ '~ * 4) bD Turb204 NGV Exit Mach number (Tip and Hub) % 1 .1 -#I E ..... : .............. 0. . . .. .. . .. ... .......... . .......... ......... €3 'C ,8 . ......... .......... .......... ................ . . ...... ... . . . . . . ........... . . . . . . ... . . . . . .. . . . . . . . . . 0.7.... .... .... 2 0.9 . ............. ................................................. ... ....... 0.8 . 0.7 . ,6 .... . .......... .......... '.......... ........... .......... .......... •.......... •.......... 0.6 . 0.5 2100 2200 2300 2400 2500 2600 time (ms) 2700 2800 Figure 6-14: NGV exit Mach number, test 204. 124 2900 3000 Turb205 NGV Exit Mach number (Tip and Hub) -1.1 -D E zi c 0 OU 0.5 2100 i I 2200 2300 I l 2400 2500 2600 time (ms) 2700 2800 Figure 6-15: NGV exit Mach number, test 205. 125 2900 3000 Synchronisation signals solid : laser pulse dash :frame transfer 2 .. ..... ... ... ..... .. o 2 . . ... ........ I . .. . . J:.. ... ... ... ."1 Il i I II . I II I ...1. 4 .... I I I I I i .. I I 1 I I i . . .I . . . . . . . . ... .. ........ I I I II I I, eIII I I I I -1 1500 2000 2500 3000 time (ms) Figure 6-16: Laser pulse and frame grabbing sync signals, test 204. 126 Synchronisation signals solid : laser pulse dash : frame transfer -11500 2500 2000 3000 time (ms) Figure 6-17: Laser pulse and frame grabbing sync signals, test 205. 127 Particle image light function Test 204 300 250 pixels pixels Figure 6-18: Particle image light intensity function, test 204. 128 Particle image contour Test 204 1 1 2 3 5 4 6 pixels Figure 6-19: Particle image contour, test 204. 129 7 Particle image light function Test 205 150- 100 50, 0 8 6 8 6 4 2 pixels 2 0 0 4 pixels pFigure Particle 6-20: image light intensity function, test 205.xels Figure 6-20: Particle image light intensity function, test 205. 130 Particle image contour Test 205 1 2 3 4 5 6 pixels Figure 6-21: Particle image contour, test 205. 131 7 Particle pair Test 204 1 4 . ....... ................ I ................................. I ...... . I I ................................. .......- 14........ ........................................... 12- - 10 6- .. 8........ ~.. ............ ..... .... . . .............. ... . ........ ............. . . •. . .................. ............... ............ ....... 2................. . ...... ....... I I I I I I 2 4 6 8 10 12 pixels Figure 6-22: Particle pair from image#3, test 204. 132 Particle pair Test 205 18 16 14 12 .x - 10 8 2 4 6 8 10 pixels 12 14 16 Figure 6-23: Particle pair from image#3, test 205. 133 18 Bibliography [1] Shang T., "Influence of Inlet Temperature Distortion on Turbine Heat Transfer" Ph.D. Thesis. Department of Aeronautics and Astronautics, Massachusetts Institute of Technology, 1995. [2] Sujudi D.D., "An Experimental Investigation of the Effects of Inlet Circumferential Temperature Distortion on the Aerodynamic Performance of a Single Stage Turbine" M.S. Thesis. Department of Aeronautics and Astronautics, Massachusetts Institute of Technology, 1993. [3] Kerrebrock J.L., "Aircraft Engines and Gas Turbines", MIT Press, 1992. [4] Harman R.T.C., "Gas Turbine Engineering", Halsted Press, 1981. [5] Bathie W.W., "Fundamentals of Gas Turbines", John Wiley and Sons, 1984. [6] Epstein A.H., Guenette, G.R., and Norton R.J.G., "The Design of the MIT Blowdown Turbine Facility", Report 183 Gas Turbine Laboratory, Department of Aeronautics and Astronautics, Massachusetts Institute of Technology, 1985. [7] Guenette G.R., "A Fully Scaled Short Duration Turbine Experiment" Sc.D. Thesis. Department of Aeronautics and Astronautics, Massachusetts Institute of Technology, 1985. [8] Haldeman C.W., "An Experimental Study of Radial Temperature Profile Effects on Turbine Tip Shroud Heat Transfer" M.S. Thesis. Department of Aeronautics and Astronautics, Massachusetts Institute of Technology, 1989. 134 [9] Graf M.B., "Investigation of the Effects of Inlet Radial Temperature Distortion on Turbine Heat Transfer" M.S. Thesis. Department of Mechanical Engineering, Massachusetts Institute of Technology, 1993. [10] Moss R.W., Oldfield M.L.G., "Measurements of Hot Combustor Turbulence Spectra", Rolls Royce, plc, 1990. [11] Perry A.E., "Hot Wire Anemometry", Oxford University Press, 1982. [12] Fried E., Idelchik I.E., "Flow Resistance", Hemisphere publishing corporation [13] "Fluid Meters, their Theory and Application", Report of ASME Research Committee on Fluid Meters, 6th edition, 1971. [14] Youngren H.H., "Analysis and Design of Transonic Cascades with Splitter Vanes" M.S. Thesis. Department of Aeronautics and Astronautics, Massachusetts Institute of Technology, 1991. [15] Udrea D.D., Bryanston-Cross P.J., Funes-Gallanzi M., Lee W.K. "High accuracy processing algorithms for particle conter estimation in low seeding density PIV", Department of Engineering, University of Warwick, UK [16] Udrea D.D., Bryanston-Cross P.J., Funes-Gallanzi M. "Advances of the Warwick Real-Time Digital 3DPIV System for Turbomachinery Reseqrch", Department of Engineering, University of Warwick, UK [17] Bryanston-Cross P.J., Private Communications. [18] Borne M., Wolfe E. "Principles of Optics" Pergamon Press, 1970. [19] van de Hulst H.C., "Light Scattering by Small Particles" Dover Publications, 1957. [20] Melling A., "Seeding Gas Flows for Laser Anemometry", AGARD. [21] Drain L.E., "The Laser Doppler Technique" Wiley, 1980. [22] Ceyhan I., Private Communications. 135